diff --git a/Attention_decode_abs_diff.png b/Attention_decode_abs_diff.png deleted file mode 100644 index 85b02ea56b030b136e1eaba5f5b0be1c51ed5356..0000000000000000000000000000000000000000 --- a/Attention_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:edc88298726e4b768cd6aead2090c71ec0047b981c68bd807675046956a430e8 -size 40543 diff --git a/Attention_decode_decode_1_H100.png b/Attention_decode_decode_1_H100.png deleted file mode 100644 index 80e3fc083115ab4f8f6648cd968091f6e6b00f8b..0000000000000000000000000000000000000000 --- a/Attention_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:1942d857c5da76a1b274190f15068039ce7ad59b58cc6eaf799182c4557da0ab -size 38763 diff --git a/Attention_decode_decode_1_MI300x.png b/Attention_decode_decode_1_MI300x.png deleted file mode 100644 index 0bb0e675a490fb0c07e302199d82fd5ee7796995..0000000000000000000000000000000000000000 --- a/Attention_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:478cd58f8a97b8539446a970687a43944cab1275a23b8219c50dfefbd5099a98 -size 39370 diff --git a/Attention_prefill_abs_diff.png b/Attention_prefill_abs_diff.png deleted file mode 100644 index 125782d5d9235a1c010600c59710fc173833aa97..0000000000000000000000000000000000000000 --- a/Attention_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:e3e067f6f85026e3e3a91ac2e53d45347d32a058df5ecbb8ec4afc1edd326646 -size 39556 diff --git a/Attention_prefill_prefill_H100.png b/Attention_prefill_prefill_H100.png deleted file mode 100644 index 495874fc49580b29af5072dbea54cc5f32f456c9..0000000000000000000000000000000000000000 --- a/Attention_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:2cbc255469cf6e91b8bd7e12a5824a354e1f84842e0e2f265752878999da6610 -size 40146 diff --git a/Attention_prefill_prefill_MI300x.png b/Attention_prefill_prefill_MI300x.png deleted file mode 100644 index 18c8e6a15f0eb2068fe19da34a706da5f8b1588c..0000000000000000000000000000000000000000 --- a/Attention_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:378fe64a9d1812e961f147984328106cfb3f127325a901a29fb72bf8fd147a5e -size 37349 diff --git a/H100_llama8b_pp1_tp1/profiling_bs1_pl4096.json b/H100_llama8b_pp1_tp1/profiling_bs1_pl4096.json deleted file mode 100644 index 379c4a85cb0b450366db78427276c07af092faf3..0000000000000000000000000000000000000000 --- a/H100_llama8b_pp1_tp1/profiling_bs1_pl4096.json +++ /dev/null @@ -1,18548 +0,0 @@ -{ - "context": { - "python_version": "3.12.9 | packaged by Anaconda, Inc. | (main, Feb 6 2025, 18:56:27) [GCC 11.2.0]", - "torch_version": "2.5.1+cu124", - "engine_args": { - "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "served_model_name": null, - "tokenizer": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "task": "auto", - "skip_tokenizer_init": false, - "tokenizer_mode": "auto", - "trust_remote_code": false, - "allowed_local_media_path": null, - "download_dir": null, - "load_format": "dummy", - "config_format": "auto", - "dtype": "auto", - "kv_cache_dtype": "auto", - "seed": 0, - "max_model_len": null, - "distributed_executor_backend": null, - "pipeline_parallel_size": 1, - "tensor_parallel_size": 1, - "max_parallel_loading_workers": null, - "block_size": null, - "enable_prefix_caching": false, - "disable_sliding_window": false, - "use_v2_block_manager": true, - "swap_space": 4, - "cpu_offload_gb": 0, - "gpu_memory_utilization": 0.9, - "max_num_batched_tokens": 8000, - "max_num_partial_prefills": 1, - "max_long_partial_prefills": 1, - "long_prefill_token_threshold": 0, - "max_num_seqs": 256, - "max_logprobs": 20, - "disable_log_stats": false, - "revision": null, - "code_revision": null, - "rope_scaling": null, - "rope_theta": null, - "hf_overrides": null, - "tokenizer_revision": null, - "quantization": null, - "enforce_eager": true, - "max_seq_len_to_capture": 8192, - "disable_custom_all_reduce": false, - "tokenizer_pool_size": 0, - "tokenizer_pool_type": "ray", - "tokenizer_pool_extra_config": null, - "limit_mm_per_prompt": null, - "mm_processor_kwargs": null, - "disable_mm_preprocessor_cache": false, - "enable_lora": false, - "enable_lora_bias": false, - "max_loras": 1, - "max_lora_rank": 16, - "enable_prompt_adapter": false, - "max_prompt_adapters": 1, - "max_prompt_adapter_token": 0, - "fully_sharded_loras": false, - "lora_extra_vocab_size": 256, - "long_lora_scaling_factors": null, - "lora_dtype": "auto", - "max_cpu_loras": null, - "device": "auto", - "num_scheduler_steps": 1, - "multi_step_stream_outputs": true, - "ray_workers_use_nsight": false, - "num_gpu_blocks_override": null, - "num_lookahead_slots": 0, - "model_loader_extra_config": null, - "ignore_patterns": [], - "preemption_mode": null, - "scheduler_delay_factor": 0.0, - "enable_chunked_prefill": null, - "guided_decoding_backend": "xgrammar", - "logits_processor_pattern": null, - "speculative_model": null, - "speculative_model_quantization": null, - "speculative_draft_tensor_parallel_size": null, - "num_speculative_tokens": null, - "speculative_disable_mqa_scorer": false, - "speculative_max_model_len": null, - "speculative_disable_by_batch_size": null, - "ngram_prompt_lookup_max": null, - "ngram_prompt_lookup_min": null, - "spec_decoding_acceptance_method": "rejection_sampler", - "typical_acceptance_sampler_posterior_threshold": null, - "typical_acceptance_sampler_posterior_alpha": null, - "qlora_adapter_name_or_path": null, - "disable_logprobs_during_spec_decoding": null, - "otlp_traces_endpoint": null, - "collect_detailed_traces": null, - "disable_async_output_proc": false, - "scheduling_policy": "fcfs", - "scheduler_cls": "vllm.core.scheduler.Scheduler", - "override_neuron_config": null, - "override_pooler_config": null, - "compilation_config": null, - "worker_cls": "auto", - "kv_transfer_config": null, - "generation_config": null, - "override_generation_config": null, - "enable_sleep_mode": false, - "model_impl": "auto", - "calculate_kv_scales": false, - "additional_config": null - }, - "prompt_len": 0, - "batch_size": 1, - "num_steps": 2, - "complete_num_requests_per_step": null, - "save_chrome_traces_folder": null - }, - "prefill": { - "metadata": { - "num_running_seqs": null - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 94699.17500000002, - "pct_cuda_time": 99.4739494416884, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 127.775, - "pct_cuda_time": 0.13421747222097483, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cuda_time_us": 127.775, - "pct_cuda_time": 0.13421747222097483, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 94525.993, - "pct_cuda_time": 99.29203552837066, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 2845.9129999999996, - "pct_cuda_time": 2.9894051967975823, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 71.264, - "pct_cuda_time": 0.07485716251501115, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 2774.6489999999994, - "pct_cuda_time": 2.9145480342825705, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 25478.952999999998, - "pct_cuda_time": 26.763613120696718, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 8821.841, - "pct_cuda_time": 9.2666421393493, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 29.729000000000013, - "pct_cuda_time": 0.031227949377087554, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 8792.112000000001, - "pct_cuda_time": 9.235414189972213, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 1748.3490000000002, - "pct_cuda_time": 1.83650153269473, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 1748.3490000000002, - "pct_cuda_time": 1.83650153269473, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 8695.786, - "pct_cuda_time": 9.134231390291855, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 729.1479999999999, - "pct_cuda_time": 0.7659119658382261, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 7918.416000000002, - "pct_cuda_time": 8.317666049807261, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 48.22200000000001, - "pct_cuda_time": 0.05065337464636939, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 6212.977, - "pct_cuda_time": 6.526238058360834, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 31.231000000000012, - "pct_cuda_time": 0.03280568088384477, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 6181.746000000001, - "pct_cuda_time": 6.493432377476989, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 66201.12700000001, - "pct_cuda_time": 69.53901721087637, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 41526.33599999999, - "pct_cuda_time": 43.62011229519753, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 31.39000000000001, - "pct_cuda_time": 0.03297269773442693, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 41494.94599999999, - "pct_cuda_time": 43.58713959746311, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 5641.842000000001, - "pct_cuda_time": 5.926306178126621, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 5641.842000000001, - "pct_cuda_time": 5.926306178126621, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 19032.949, - "pct_cuda_time": 19.992598737552186, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 32.254000000000005, - "pct_cuda_time": 0.0338802609979677, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 19000.695000000003, - "pct_cuda_time": 19.95871847655422, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 45.407, - "pct_cuda_time": 0.04769644109675447, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 45.407, - "pct_cuda_time": 0.04769644109675447, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 356.608, - "pct_cuda_time": 0.3745883336629027, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.265, - "pct_cuda_time": 0.0034296227493757214, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 352.575, - "pct_cuda_time": 0.37035198801260183, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 144.192, - "pct_cuda_time": 0.1514622246486934, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 16.544, - "pct_cuda_time": 0.017378155824095538, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.704, - "pct_cuda_time": 0.0049411777681664295, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 5.376, - "pct_cuda_time": 0.00564706030647592, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 40.64, - "pct_cuda_time": 0.04268908684062153, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 32.576, - "pct_cuda_time": 0.034218496380907654, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 2.144, - "pct_cuda_time": 0.0022521014317493254, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 5.312, - "pct_cuda_time": 0.005579833398065492, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 33.728, - "pct_cuda_time": 0.03542858073229535, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 3.168, - "pct_cuda_time": 0.003327731966316167, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 17255.819, - "cuda_time_us": 94699.17500000002, - "pct_cuda_time": 99.4739494416884, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 66.087, - "cuda_time_us": 127.775, - "pct_cuda_time": 0.13421747222097483, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 127.775, - "pct_cuda_time": 0.13421747222097483, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[4096]) <- embedding(bfloat16[128256, 4096], int64[4096], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 865.73, - "cuda_time_us": 3222.3269999999998, - "pct_cuda_time": 3.3847981577726247, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 48.726, - "cuda_time_us": 71.264, - "pct_cuda_time": 0.07485716251501115, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 71.264, - "pct_cuda_time": 0.07485716251501115, - "trace": "_C::rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 627.754, - "cuda_time_us": 878.684, - "pct_cuda_time": 0.9229876373391902, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 87.367, - "cuda_time_us": 300.991, - "pct_cuda_time": 0.31616709983379715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 300.255, - "pct_cuda_time": 0.3153939903870772, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 194.088, - "cuda_time_us": 57.152, - "pct_cuda_time": 0.060033629210511856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.152, - "pct_cuda_time": 0.060033629210511856, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 217.039, - "cuda_time_us": 305.919, - "pct_cuda_time": 0.32134357178140005, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.52, - "pct_cuda_time": 0.02470588884083215, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 280.799, - "pct_cuda_time": 0.29495701023030724, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.0016806727102606906, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 73.253, - "cuda_time_us": 214.622, - "pct_cuda_time": 0.2254433365134812, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007720590262760047, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 213.887, - "pct_cuda_time": 0.22467127748720517, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.61, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.046252112986374196, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.046252112986374196, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 128.122, - "cuda_time_us": 2228.3469999999998, - "pct_cuda_time": 2.340701244932049, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 44.016, - "cuda_time_us": 1419.037, - "pct_cuda_time": 1.4905854754688745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1418.269, - "pct_cuda_time": 1.4897787525679493, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 29.454, - "cuda_time_us": 182.079, - "pct_cuda_time": 0.19125950400722266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 182.079, - "pct_cuda_time": 0.19125950400722266, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 38.651, - "cuda_time_us": 627.231, - "pct_cuda_time": 0.658856265455952, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 626.495, - "pct_cuda_time": 0.6580831560092321, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 582.613, - "cuda_time_us": 2951.322, - "pct_cuda_time": 3.100128965370001, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.859, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.046218499532168986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.046218499532168986, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 421.202, - "cuda_time_us": 815.1670000000001, - "pct_cuda_time": 0.8562680820031728, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 35.504, - "cuda_time_us": 278.975, - "pct_cuda_time": 0.2930410433406101, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 277.951, - "pct_cuda_time": 0.2919654128060432, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 125.134, - "cuda_time_us": 54.977, - "pct_cuda_time": 0.05774896474500123, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.977, - "pct_cuda_time": 0.05774896474500123, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 178.446, - "cuda_time_us": 282.68800000000005, - "pct_cuda_time": 0.2969412544488588, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.648, - "pct_cuda_time": 0.024840342657653003, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 257.696, - "pct_cuda_time": 0.2706891467145868, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.00141176507661898, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 46.477, - "cuda_time_us": 198.52700000000002, - "pct_cuda_time": 0.2085368194687026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011764708971824835, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 197.407, - "pct_cuda_time": 0.20736034857152008, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.452, - "cuda_time_us": 44.639, - "pct_cuda_time": 0.04688971819582935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.639, - "pct_cuda_time": 0.04688971819582935, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.658, - "cuda_time_us": 2047.516, - "pct_cuda_time": 2.15075266563883, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.735, - "cuda_time_us": 1292.1570000000002, - "pct_cuda_time": 1.3573081295452019, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.00141176507661898, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1290.813, - "pct_cuda_time": 1.3558963644685829, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 23.348, - "cuda_time_us": 174.304, - "pct_cuda_time": 0.18309248505579961, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.304, - "pct_cuda_time": 0.18309248505579961, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.55, - "cuda_time_us": 581.055, - "pct_cuda_time": 0.6103520510378283, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 580.319, - "pct_cuda_time": 0.6095789415911084, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 523.093, - "cuda_time_us": 2890.967, - "pct_cuda_time": 3.0367308394776362, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.391, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.047159676249914975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.047159676249914975, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 362.922, - "cuda_time_us": 780.702, - "pct_cuda_time": 0.8200653414037135, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.708, - "cuda_time_us": 271.07099999999997, - "pct_cuda_time": 0.28473852015192225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.0014453785308241936, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.695, - "pct_cuda_time": 0.28329314162109803, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.264, - "cuda_time_us": 54.016, - "pct_cuda_time": 0.056739510698400906, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.016, - "pct_cuda_time": 0.056739510698400906, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.065, - "cuda_time_us": 266.527, - "pct_cuda_time": 0.2799654096547819, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.4, - "pct_cuda_time": 0.023529417943649662, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 242.559, - "pct_cuda_time": 0.2547889324550768, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.568, - "pct_cuda_time": 0.0016470592560554765, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.246, - "cuda_time_us": 189.088, - "pct_cuda_time": 0.1986219008986084, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.32, - "pct_cuda_time": 0.19781517799768322, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.057, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.045512616993859493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.045512616993859493, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.961, - "cuda_time_us": 2022.0410000000002, - "pct_cuda_time": 2.123993204830148, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.337, - "cuda_time_us": 1263.613, - "pct_cuda_time": 1.3273249283941513, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1262.845, - "pct_cuda_time": 1.326518205493226, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.046, - "cuda_time_us": 175.007, - "pct_cuda_time": 0.1838309306278704, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 175.007, - "pct_cuda_time": 0.1838309306278704, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.185, - "cuda_time_us": 583.4209999999999, - "pct_cuda_time": 0.6128373458081263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.087, - "pct_cuda_time": 0.0011418070225333564, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.334, - "pct_cuda_time": 0.611695538785593, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 528.428, - "cuda_time_us": 2891.352, - "pct_cuda_time": 3.037135251348542, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.071, - "cuda_time_us": 44.704, - "pct_cuda_time": 0.04695799552468369, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.704, - "pct_cuda_time": 0.04695799552468369, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 376.622, - "cuda_time_us": 783.6759999999999, - "pct_cuda_time": 0.8231892918039104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.857, - "cuda_time_us": 271.39099999999996, - "pct_cuda_time": 0.28507465469397436, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 270.655, - "pct_cuda_time": 0.28430154524725443, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.531, - "cuda_time_us": 53.92, - "pct_cuda_time": 0.05663867033578526, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.92, - "pct_cuda_time": 0.05663867033578526, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 162.877, - "cuda_time_us": 266.654, - "pct_cuda_time": 0.28009881305115886, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.655, - "pct_cuda_time": 0.023797275156847465, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 242.368, - "pct_cuda_time": 0.2545883021502894, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.631, - "pct_cuda_time": 0.0017132357440219914, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.676, - "cuda_time_us": 191.711, - "pct_cuda_time": 0.20137715372299203, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.728, - "pct_cuda_time": 0.0018151265270815455, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.983, - "pct_cuda_time": 0.19956202719591046, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.674, - "cuda_time_us": 42.687, - "pct_cuda_time": 0.04483929748931131, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 42.687, - "pct_cuda_time": 0.04483929748931131, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 103.028, - "cuda_time_us": 2020.285, - "pct_cuda_time": 2.122148666530637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.575, - "cuda_time_us": 1262.622, - "pct_cuda_time": 1.3262839617342335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1261.854, - "pct_cuda_time": 1.3254772388333083, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.474, - "cuda_time_us": 175.104, - "pct_cuda_time": 0.18393282141092995, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 175.104, - "pct_cuda_time": 0.18393282141092995, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.154, - "cuda_time_us": 582.559, - "pct_cuda_time": 0.6119318833854734, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0014789919850294075, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.151, - "pct_cuda_time": 0.6104528914004439, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 531.975, - "cuda_time_us": 2886.232, - "pct_cuda_time": 3.0317570986757083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.618, - "cuda_time_us": 43.872, - "pct_cuda_time": 0.04608404571534813, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.872, - "pct_cuda_time": 0.04608404571534813, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 376.051, - "cuda_time_us": 781.151, - "pct_cuda_time": 0.8205369801830302, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.781, - "cuda_time_us": 270.751, - "pct_cuda_time": 0.2844023856098701, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0010084036261564142, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.791, - "pct_cuda_time": 0.2833939819837137, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.423, - "cuda_time_us": 54.048, - "pct_cuda_time": 0.05677312415260612, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.048, - "pct_cuda_time": 0.05677312415260612, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.396, - "cuda_time_us": 266.848, - "pct_cuda_time": 0.28030259461727797, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.368, - "pct_cuda_time": 0.023495804489444452, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 242.816, - "pct_cuda_time": 0.25505889050916236, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.001747899618671118, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.869, - "cuda_time_us": 189.504, - "pct_cuda_time": 0.19905887580327616, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.768, - "pct_cuda_time": 0.19828576635655626, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.506, - "cuda_time_us": 43.455, - "pct_cuda_time": 0.045646020390236436, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.455, - "pct_cuda_time": 0.045646020390236436, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.487, - "cuda_time_us": 2017.754, - "pct_cuda_time": 2.119490052387093, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.687, - "cuda_time_us": 1261.596, - "pct_cuda_time": 1.3252062303587786, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011092439887720558, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1260.54, - "pct_cuda_time": 1.3240969863700067, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.84, - "cuda_time_us": 174.815, - "pct_cuda_time": 0.18362924990263912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.815, - "pct_cuda_time": 0.18362924990263912, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 34.023, - "cuda_time_us": 581.343, - "pct_cuda_time": 0.6106545721256753, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 580.607, - "pct_cuda_time": 0.6098814626789554, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 523.021, - "cuda_time_us": 2889.978, - "pct_cuda_time": 3.035691973658606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.966, - "cuda_time_us": 43.776, - "pct_cuda_time": 0.04598320535273249, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.776, - "pct_cuda_time": 0.04598320535273249, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 373.991, - "cuda_time_us": 779.9669999999999, - "pct_cuda_time": 0.8192932823774374, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.831, - "cuda_time_us": 270.143, - "pct_cuda_time": 0.283763729979971, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.407, - "pct_cuda_time": 0.2829906205332511, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 113.59, - "cuda_time_us": 53.664, - "pct_cuda_time": 0.056369762702143555, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.664, - "pct_cuda_time": 0.056369762702143555, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.238, - "cuda_time_us": 265.823, - "pct_cuda_time": 0.2792259136622672, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.008, - "pct_cuda_time": 0.024168073573548728, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 241.471, - "pct_cuda_time": 0.25364607501209946, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.00141176507661898, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.029, - "cuda_time_us": 190.337, - "pct_cuda_time": 0.19993387603305562, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0011102944092159686, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.28, - "pct_cuda_time": 0.19882358162383967, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.166, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.045647070810680354, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.045647070810680354, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.791, - "cuda_time_us": 2022.779, - "pct_cuda_time": 2.1247684151177557, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.875, - "cuda_time_us": 1266.333, - "pct_cuda_time": 1.3301820720015944, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.001747899618671118, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1264.669, - "pct_cuda_time": 1.3284341723829232, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.15, - "cuda_time_us": 174.752, - "pct_cuda_time": 0.1835630734146726, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.752, - "pct_cuda_time": 0.1835630734146726, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.702, - "cuda_time_us": 581.694, - "pct_cuda_time": 0.6110232697014887, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011428574429772696, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 580.606, - "pct_cuda_time": 0.6098804122585115, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 526.001, - "cuda_time_us": 2887.9970000000003, - "pct_cuda_time": 3.033611090759215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.739, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.04594959189852727, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.04594959189852727, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 381.098, - "cuda_time_us": 779.7440000000001, - "pct_cuda_time": 0.819059038618445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.51, - "cuda_time_us": 271.264, - "pct_cuda_time": 0.28494125129759745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.0013781516224137661, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.952, - "pct_cuda_time": 0.28356309967518367, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 110.113, - "cuda_time_us": 53.919, - "pct_cuda_time": 0.05663761991534135, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.919, - "pct_cuda_time": 0.05663761991534135, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 166.246, - "cuda_time_us": 265.40900000000005, - "pct_cuda_time": 0.2787910395984873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.008, - "pct_cuda_time": 0.024168073573548728, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 241.056, - "pct_cuda_time": 0.2532101505278756, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.345, - "pct_cuda_time": 0.0014128154970628928, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.999, - "cuda_time_us": 189.152, - "pct_cuda_time": 0.1986891278070188, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.416, - "pct_cuda_time": 0.1979160183602989, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.82, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.04578152462750121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.04578152462750121, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.561, - "cuda_time_us": 2020.925, - "pct_cuda_time": 2.122820935614741, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.333, - "cuda_time_us": 1263.006, - "pct_cuda_time": 1.326687323184696, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0011102944092159686, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1261.949, - "pct_cuda_time": 1.32557702877548, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.256, - "cuda_time_us": 175.072, - "pct_cuda_time": 0.18389920795672474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 175.072, - "pct_cuda_time": 0.18389920795672474, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.836, - "cuda_time_us": 582.847, - "pct_cuda_time": 0.6122344044733203, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.0014453785308241936, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.471, - "pct_cuda_time": 0.6107890259424962, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 522.652, - "cuda_time_us": 2888.0899999999997, - "pct_cuda_time": 3.033708779860498, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.747, - "cuda_time_us": 44.288, - "pct_cuda_time": 0.04652102062001591, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.288, - "pct_cuda_time": 0.04652102062001591, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 371.043, - "cuda_time_us": 779.198, - "pct_cuda_time": 0.8184855090560683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.005, - "cuda_time_us": 271.135, - "pct_cuda_time": 0.2848057470603326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 270.399, - "pct_cuda_time": 0.28403263761361275, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 107.188, - "cuda_time_us": 54.112, - "pct_cuda_time": 0.05684035106101654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.112, - "pct_cuda_time": 0.05684035106101654, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.677, - "cuda_time_us": 264.127, - "pct_cuda_time": 0.27744440058939085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.24, - "pct_cuda_time": 0.023361350672623595, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.383, - "pct_cuda_time": 0.2525032175691222, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.001579832347645049, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 45.195, - "cuda_time_us": 189.82399999999998, - "pct_cuda_time": 0.19939501034532828, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0014789919850294075, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.416, - "pct_cuda_time": 0.1979160183602989, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.662, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.04631933989478462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.04631933989478462, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.544, - "cuda_time_us": 2020.5079999999998, - "pct_cuda_time": 2.122382910289629, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.685, - "cuda_time_us": 1261.981, - "pct_cuda_time": 1.3256106422296852, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1261.213, - "pct_cuda_time": 1.3248039193287602, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.57, - "cuda_time_us": 174.432, - "pct_cuda_time": 0.18322693887262045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.432, - "pct_cuda_time": 0.18322693887262045, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.653, - "cuda_time_us": 584.0949999999999, - "pct_cuda_time": 0.6135453291873236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0016134458018502626, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.559, - "pct_cuda_time": 0.6119318833854734, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 524.462, - "cuda_time_us": 2886.202, - "pct_cuda_time": 3.031725586062391, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.254, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.04591597844432207, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.04591597844432207, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 369.849, - "cuda_time_us": 778.687, - "pct_cuda_time": 0.8179487442092289, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.606, - "cuda_time_us": 270.079, - "pct_cuda_time": 0.28369650307156064, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.343, - "pct_cuda_time": 0.28292339362484076, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.906, - "cuda_time_us": 54.432, - "pct_cuda_time": 0.057176485603068684, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.432, - "pct_cuda_time": 0.057176485603068684, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.013, - "cuda_time_us": 264.512, - "pct_cuda_time": 0.27784881246029736, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.4, - "pct_cuda_time": 0.023529417943649662, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.384, - "pct_cuda_time": 0.2525042679895661, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.728, - "pct_cuda_time": 0.0018151265270815455, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.025, - "cuda_time_us": 189.664, - "pct_cuda_time": 0.1992269430743022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.928, - "pct_cuda_time": 0.19845383362758232, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.634, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.04658824752842633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.04658824752842633, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 106.93, - "cuda_time_us": 2019.451, - "pct_cuda_time": 2.1212726158804136, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.64, - "cuda_time_us": 1263.325, - "pct_cuda_time": 1.3270224073063042, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011428574429772696, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1262.237, - "pct_cuda_time": 1.3258795498633271, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.815, - "cuda_time_us": 174.528, - "pct_cuda_time": 0.1833277792352361, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.528, - "pct_cuda_time": 0.1833277792352361, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 37.964, - "cuda_time_us": 581.598, - "pct_cuda_time": 0.6109224293388731, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 580.862, - "pct_cuda_time": 0.6101493198921532, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 519.703, - "cuda_time_us": 2888.6989999999996, - "pct_cuda_time": 3.034348485910841, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.059, - "cuda_time_us": 44.193, - "pct_cuda_time": 0.04642123067784418, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.193, - "pct_cuda_time": 0.04642123067784418, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 367.633, - "cuda_time_us": 780.255, - "pct_cuda_time": 0.8195958034652843, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.514, - "cuda_time_us": 270.432, - "pct_cuda_time": 0.2840673014882619, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.408, - "pct_cuda_time": 0.28299167095369504, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.285, - "cuda_time_us": 54.368, - "pct_cuda_time": 0.057109258694658264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.368, - "pct_cuda_time": 0.057109258694658264, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.515, - "cuda_time_us": 264.575, - "pct_cuda_time": 0.27791498894826383, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.528, - "pct_cuda_time": 0.02366387176047052, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.576, - "pct_cuda_time": 0.2527059487147974, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.471, - "pct_cuda_time": 0.0015451684729959224, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.42, - "cuda_time_us": 190.88, - "pct_cuda_time": 0.20050425433410035, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.856, - "pct_cuda_time": 0.19942862379953352, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.138, - "cuda_time_us": 43.84, - "pct_cuda_time": 0.04605043226114292, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.84, - "pct_cuda_time": 0.04605043226114292, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.238, - "cuda_time_us": 2020.4109999999996, - "pct_cuda_time": 2.1222810195065693, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.5, - "cuda_time_us": 1263.3249999999998, - "pct_cuda_time": 1.327022407306304, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.001578781927201136, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1261.822, - "pct_cuda_time": 1.3254436253791029, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.382, - "cuda_time_us": 174.591, - "pct_cuda_time": 0.18339395572320263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.591, - "pct_cuda_time": 0.18339395572320263, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.58, - "cuda_time_us": 582.495, - "pct_cuda_time": 0.611864656477063, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011092439887720558, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.439, - "pct_cuda_time": 0.6107554124882909, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 518.174, - "cuda_time_us": 2889.498, - "pct_cuda_time": 3.0351877718455276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.858, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.04658824752842633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.04658824752842633, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 373.315, - "cuda_time_us": 778.111, - "pct_cuda_time": 0.817343702033535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.84, - "cuda_time_us": 269.087, - "pct_cuda_time": 0.282654485991199, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 268.351, - "pct_cuda_time": 0.2818813765444791, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 115.585, - "cuda_time_us": 54.208, - "pct_cuda_time": 0.05694119142363219, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.208, - "pct_cuda_time": 0.05694119142363219, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.912, - "cuda_time_us": 264.32, - "pct_cuda_time": 0.27764713173506606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.464, - "pct_cuda_time": 0.023596644852060093, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.512, - "pct_cuda_time": 0.25263872180638697, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.00141176507661898, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.914, - "cuda_time_us": 190.49599999999998, - "pct_cuda_time": 0.2001008928836378, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.76, - "pct_cuda_time": 0.19932778343691787, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.345, - "cuda_time_us": 44.128, - "pct_cuda_time": 0.04635295334898984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.128, - "pct_cuda_time": 0.04635295334898984, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.913, - "cuda_time_us": 2022.907, - "pct_cuda_time": 2.1249028689345764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.063, - "cuda_time_us": 1266.2369999999999, - "pct_cuda_time": 1.3300812316389785, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1265.213, - "pct_cuda_time": 1.3290056011044116, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.678, - "cuda_time_us": 174.719, - "pct_cuda_time": 0.18352840954002347, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.719, - "pct_cuda_time": 0.18352840954002347, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.026, - "cuda_time_us": 581.951, - "pct_cuda_time": 0.6112932277555744, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.215, - "pct_cuda_time": 0.6105201183088546, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 514.779, - "cuda_time_us": 2886.105, - "pct_cuda_time": 3.0316236952793307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.004, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.04611765916955335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.04611765916955335, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 366.733, - "cuda_time_us": 779.39, - "pct_cuda_time": 0.8186871897812996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.587, - "cuda_time_us": 270.368, - "pct_cuda_time": 0.28400007457985144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.344, - "pct_cuda_time": 0.2829244440452846, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.786, - "cuda_time_us": 54.4, - "pct_cuda_time": 0.05714287214886347, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.4, - "pct_cuda_time": 0.05714287214886347, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.201, - "cuda_time_us": 264.959, - "pct_cuda_time": 0.27831835039872643, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.4, - "pct_cuda_time": 0.023529417943649662, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.863, - "pct_cuda_time": 0.25300741938220045, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.696, - "pct_cuda_time": 0.0017815130728763317, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.511, - "cuda_time_us": 189.66299999999998, - "pct_cuda_time": 0.19922589265385832, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.927, - "pct_cuda_time": 0.19845278320713838, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.895, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.045848751535911635, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.045848751535911635, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.709, - "cuda_time_us": 2019.163, - "pct_cuda_time": 2.1209700947925665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.532, - "cuda_time_us": 1262.461, - "pct_cuda_time": 1.3261148440427635, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1261.693, - "pct_cuda_time": 1.3253081211418383, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.834, - "cuda_time_us": 174.368, - "pct_cuda_time": 0.18315971196421005, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.368, - "pct_cuda_time": 0.18315971196421005, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.864, - "cuda_time_us": 582.3340000000001, - "pct_cuda_time": 0.6116955387855931, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.0013781516224137661, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.022, - "pct_cuda_time": 0.6103173871631794, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 534.281, - "cuda_time_us": 2889.374, - "pct_cuda_time": 3.0350575197104823, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.113, - "cuda_time_us": 44.641, - "pct_cuda_time": 0.04689181903671717, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.641, - "pct_cuda_time": 0.04689181903671717, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 389.037, - "cuda_time_us": 779.5830000000001, - "pct_cuda_time": 0.818889920926975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.563, - "cuda_time_us": 270.94500000000005, - "pct_cuda_time": 0.28460616717598924, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0007741598671638304, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 270.208, - "pct_cuda_time": 0.2838320073088254, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.628, - "cuda_time_us": 54.048, - "pct_cuda_time": 0.05677312415260612, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.048, - "pct_cuda_time": 0.05677312415260612, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 166.293, - "cuda_time_us": 265.43800000000005, - "pct_cuda_time": 0.2788215017913607, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.623, - "pct_cuda_time": 0.02376366170264225, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 241.311, - "pct_cuda_time": 0.2534780077410734, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.001579832347645049, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.82, - "cuda_time_us": 189.152, - "pct_cuda_time": 0.1986891278070188, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.416, - "pct_cuda_time": 0.1979160183602989, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.42, - "cuda_time_us": 43.2, - "pct_cuda_time": 0.04537816317703865, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.2, - "pct_cuda_time": 0.04537816317703865, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.672, - "cuda_time_us": 2021.9499999999998, - "pct_cuda_time": 2.1238976165697516, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.003, - "cuda_time_us": 1264.127, - "pct_cuda_time": 1.3278648445023222, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0011102944092159686, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1263.07, - "pct_cuda_time": 1.3267545500931062, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.313, - "cuda_time_us": 174.24, - "pct_cuda_time": 0.18302525814738918, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.24, - "pct_cuda_time": 0.18302525814738918, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.258, - "cuda_time_us": 583.583, - "pct_cuda_time": 0.6130075139200403, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.847, - "pct_cuda_time": 0.6122344044733203, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 489.202, - "cuda_time_us": 2886.265, - "pct_cuda_time": 3.0317917625503568, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.907, - "cuda_time_us": 45.376, - "pct_cuda_time": 0.04766387806299318, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.376, - "pct_cuda_time": 0.04766387806299318, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 348.722, - "cuda_time_us": 778.2379999999999, - "pct_cuda_time": 0.817477105429912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.863, - "cuda_time_us": 270.175, - "pct_cuda_time": 0.2837973434341763, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.151, - "pct_cuda_time": 0.2827217128996094, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.311, - "cuda_time_us": 54.015, - "pct_cuda_time": 0.056738460277956995, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.015, - "pct_cuda_time": 0.056738460277956995, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 144.675, - "cuda_time_us": 264.928, - "pct_cuda_time": 0.2782857873649651, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.336, - "pct_cuda_time": 0.023462191035239235, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 241.28, - "pct_cuda_time": 0.2534454447073121, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.0013781516224137661, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.548, - "cuda_time_us": 189.11999999999998, - "pct_cuda_time": 0.19865551435281356, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0011102944092159686, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.063, - "pct_cuda_time": 0.19754521994359764, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.886, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.04521009590601257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.04521009590601257, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.733, - "cuda_time_us": 2019.611, - "pct_cuda_time": 2.1214406831514396, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.556, - "cuda_time_us": 1262.621, - "pct_cuda_time": 1.3262829113137895, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1261.885, - "pct_cuda_time": 1.3255098018670695, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.202, - "cuda_time_us": 173.855, - "pct_cuda_time": 0.18262084627648267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 173.855, - "pct_cuda_time": 0.18262084627648267, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.191, - "cuda_time_us": 583.135, - "pct_cuda_time": 0.6125369255611672, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011092439887720558, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.079, - "pct_cuda_time": 0.6114276815723952, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 518.435, - "cuda_time_us": 2887.2599999999998, - "pct_cuda_time": 3.03283693089205, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.369, - "cuda_time_us": 44.8, - "pct_cuda_time": 0.047058835887299325, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.8, - "pct_cuda_time": 0.047058835887299325, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 371.191, - "cuda_time_us": 777.8549999999999, - "pct_cuda_time": 0.8170747943998932, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.771, - "cuda_time_us": 269.792, - "pct_cuda_time": 0.2833950324041576, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.00141176507661898, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 268.448, - "pct_cuda_time": 0.2819832673275386, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.088, - "cuda_time_us": 54.272, - "pct_cuda_time": 0.057008418332042614, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.272, - "pct_cuda_time": 0.057008418332042614, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.609, - "cuda_time_us": 264.159, - "pct_cuda_time": 0.27747801404359607, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.272, - "pct_cuda_time": 0.02339496412682881, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.511, - "pct_cuda_time": 0.25263767138594306, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.0014453785308241936, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.91, - "cuda_time_us": 189.63199999999998, - "pct_cuda_time": 0.199193329620097, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.896, - "pct_cuda_time": 0.19842022017337707, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.96, - "cuda_time_us": 43.392, - "pct_cuda_time": 0.04557984390226993, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.392, - "pct_cuda_time": 0.04557984390226993, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.011, - "cuda_time_us": 2021.213, - "pct_cuda_time": 2.123123456702588, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.396, - "cuda_time_us": 1263.4859999999999, - "pct_cuda_time": 1.327191524997774, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011764708971824835, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1262.366, - "pct_cuda_time": 1.3260150541005917, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.404, - "cuda_time_us": 174.688, - "pct_cuda_time": 0.18349584650626216, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.688, - "pct_cuda_time": 0.18349584650626216, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.775, - "cuda_time_us": 583.039, - "pct_cuda_time": 0.6124360851985516, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.303, - "pct_cuda_time": 0.6116629757518317, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 503.524, - "cuda_time_us": 2890.0730000000003, - "pct_cuda_time": 3.035791763600778, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.693, - "cuda_time_us": 44.544, - "pct_cuda_time": 0.04678992825365762, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.544, - "pct_cuda_time": 0.04678992825365762, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 354.327, - "cuda_time_us": 779.232, - "pct_cuda_time": 0.8185212233511614, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.0, - "cuda_time_us": 269.889, - "pct_cuda_time": 0.2834969231872172, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.993, - "pct_cuda_time": 0.001043067500805541, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 268.896, - "pct_cuda_time": 0.2824538556864117, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.924, - "cuda_time_us": 53.888, - "pct_cuda_time": 0.056605056881580046, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.888, - "pct_cuda_time": 0.056605056881580046, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 147.664, - "cuda_time_us": 264.511, - "pct_cuda_time": 0.27784776203985345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.304, - "pct_cuda_time": 0.023428577581034022, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.703, - "pct_cuda_time": 0.25283935211117436, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.001579832347645049, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.534, - "cuda_time_us": 190.944, - "pct_cuda_time": 0.20057148124251076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015126054392346213, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.504, - "pct_cuda_time": 0.19905887580327616, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.955, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.045512616993859493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.045512616993859493, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 103.164, - "cuda_time_us": 2022.9690000000003, - "pct_cuda_time": 2.124967995002099, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.35, - "cuda_time_us": 1265.4050000000002, - "pct_cuda_time": 1.3292072818296432, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1264.669, - "pct_cuda_time": 1.3284341723829232, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.654, - "cuda_time_us": 174.911, - "pct_cuda_time": 0.18373009026525475, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.911, - "pct_cuda_time": 0.18373009026525475, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 38.097, - "cuda_time_us": 582.653, - "pct_cuda_time": 0.6120306229072013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.311, - "pct_cuda_time": 0.001377101201969853, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.342, - "pct_cuda_time": 0.6106535217052315, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 500.18, - "cuda_time_us": 2885.53, - "pct_cuda_time": 3.0310197035240813, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.534, - "cuda_time_us": 44.736, - "pct_cuda_time": 0.046991608978888905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.736, - "pct_cuda_time": 0.046991608978888905, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 354.171, - "cuda_time_us": 777.311, - "pct_cuda_time": 0.8165033656784048, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.562, - "cuda_time_us": 269.59999999999997, - "pct_cuda_time": 0.2831933516789263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 268.864, - "pct_cuda_time": 0.28242024223220635, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.186, - "cuda_time_us": 53.984, - "pct_cuda_time": 0.0567058972441957, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.984, - "pct_cuda_time": 0.0567058972441957, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.066, - "cuda_time_us": 264.512, - "pct_cuda_time": 0.27784881246029736, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.56, - "pct_cuda_time": 0.023697485214675733, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.448, - "pct_cuda_time": 0.25257149489797653, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.001579832347645049, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.161, - "cuda_time_us": 189.215, - "pct_cuda_time": 0.19875530429498534, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011092439887720558, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.159, - "pct_cuda_time": 0.19764606030621326, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.784, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.045647070810680354, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.045647070810680354, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.524, - "cuda_time_us": 2020.027, - "pct_cuda_time": 2.1218776580561074, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.921, - "cuda_time_us": 1262.749, - "pct_cuda_time": 1.3264173651306104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1261.981, - "pct_cuda_time": 1.3256106422296852, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.192, - "cuda_time_us": 175.008, - "pct_cuda_time": 0.18383198104831433, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 175.008, - "pct_cuda_time": 0.18383198104831433, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.547, - "cuda_time_us": 582.27, - "pct_cuda_time": 0.6116283118771826, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.534, - "pct_cuda_time": 0.6108552024304627, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 546.694, - "cuda_time_us": 2891.098, - "pct_cuda_time": 3.0368684445557883, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.329, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.045848751535911635, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.045848751535911635, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 380.947, - "cuda_time_us": 778.5889999999999, - "pct_cuda_time": 0.8178458030057253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.909, - "cuda_time_us": 270.49600000000004, - "pct_cuda_time": 0.28413452839667236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.001042017080361628, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.504, - "pct_cuda_time": 0.2830925113163107, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 110.521, - "cuda_time_us": 54.207, - "pct_cuda_time": 0.056940141003188276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.207, - "pct_cuda_time": 0.056940141003188276, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.826, - "cuda_time_us": 264.83, - "pct_cuda_time": 0.27818284616146166, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.008, - "pct_cuda_time": 0.024168073573548728, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.351, - "pct_cuda_time": 0.252469604114917, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.471, - "pct_cuda_time": 0.0015451684729959224, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.0, - "cuda_time_us": 189.05599999999998, - "pct_cuda_time": 0.19858828744440316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.32, - "pct_cuda_time": 0.19781517799768322, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.677, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.04521009590601257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.04521009590601257, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 117.127, - "cuda_time_us": 2025.821, - "pct_cuda_time": 2.1279637941081386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.59, - "cuda_time_us": 1266.046, - "pct_cuda_time": 1.3298806013341913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015126054392346213, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1264.606, - "pct_cuda_time": 1.3283679958949568, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 24.435, - "cuda_time_us": 174.624, - "pct_cuda_time": 0.18342861959785173, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.624, - "pct_cuda_time": 0.18342861959785173, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 34.606, - "cuda_time_us": 585.151, - "pct_cuda_time": 0.6146545731760957, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011428574429772696, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 584.063, - "pct_cuda_time": 0.6135117157331185, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 498.057, - "cuda_time_us": 2888.92, - "pct_cuda_time": 3.0345806288289463, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.883, - "cuda_time_us": 44.383, - "pct_cuda_time": 0.04662081056218764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.383, - "pct_cuda_time": 0.04662081056218764, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 351.15, - "cuda_time_us": 778.7819999999999, - "pct_cuda_time": 0.8180485341514006, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.073, - "cuda_time_us": 270.143, - "pct_cuda_time": 0.283763729979971, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.407, - "pct_cuda_time": 0.2829906205332511, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.165, - "cuda_time_us": 53.472, - "pct_cuda_time": 0.056168081976912275, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.472, - "pct_cuda_time": 0.056168081976912275, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 149.239, - "cuda_time_us": 264.96, - "pct_cuda_time": 0.2783194008191703, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.976, - "pct_cuda_time": 0.02413446011934351, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.64, - "pct_cuda_time": 0.25277317562320784, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.00141176507661898, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.552, - "cuda_time_us": 190.207, - "pct_cuda_time": 0.19979732137534695, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.215, - "pct_cuda_time": 0.0012762608393542118, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.992, - "pct_cuda_time": 0.19852106053599275, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.642, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.046252112986374196, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.046252112986374196, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.386, - "cuda_time_us": 2021.723, - "pct_cuda_time": 2.1236591711289834, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.67, - "cuda_time_us": 1264.157, - "pct_cuda_time": 1.3278963571156397, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1263.389, - "pct_cuda_time": 1.3270896342147145, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.224, - "cuda_time_us": 174.688, - "pct_cuda_time": 0.18349584650626216, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.688, - "pct_cuda_time": 0.18349584650626216, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.067, - "cuda_time_us": 582.878, - "pct_cuda_time": 0.6122669675070816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.863, - "pct_cuda_time": 0.0009065128430968598, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.015, - "pct_cuda_time": 0.6113604546639848, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 518.638, - "cuda_time_us": 2889.399, - "pct_cuda_time": 3.0350837802215804, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.126, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.046016818806937705, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.046016818806937705, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 365.823, - "cuda_time_us": 779.261, - "pct_cuda_time": 0.8185516855440349, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.136, - "cuda_time_us": 270.59200000000004, - "pct_cuda_time": 0.284235368759288, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.001042017080361628, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.6, - "pct_cuda_time": 0.28319335167892634, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 96.881, - "cuda_time_us": 53.727, - "pct_cuda_time": 0.056435939190110064, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.727, - "pct_cuda_time": 0.056435939190110064, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 167.878, - "cuda_time_us": 265.279, - "pct_cuda_time": 0.27865448494077855, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.528, - "pct_cuda_time": 0.02366387176047052, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 241.088, - "pct_cuda_time": 0.2532437639820808, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.663, - "pct_cuda_time": 0.001746849198227205, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.891, - "cuda_time_us": 189.66299999999998, - "pct_cuda_time": 0.19922589265385832, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.927, - "pct_cuda_time": 0.19845278320713838, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.59, - "cuda_time_us": 44.223, - "pct_cuda_time": 0.04645274329116157, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.223, - "pct_cuda_time": 0.04645274329116157, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.711, - "cuda_time_us": 2022.107, - "pct_cuda_time": 2.1240625325794458, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.975, - "cuda_time_us": 1264.6039999999998, - "pct_cuda_time": 1.3283658950540687, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1263.58, - "pct_cuda_time": 1.327290264519502, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.167, - "cuda_time_us": 174.88, - "pct_cuda_time": 0.18369752723149346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.88, - "pct_cuda_time": 0.18369752723149346, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.257, - "cuda_time_us": 582.623, - "pct_cuda_time": 0.6119991102938839, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.0013781516224137661, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.311, - "pct_cuda_time": 0.6106209586714701, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 515.238, - "cuda_time_us": 2892.633, - "pct_cuda_time": 3.0384808399371943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.985, - "cuda_time_us": 43.647, - "pct_cuda_time": 0.045847701115467716, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.647, - "pct_cuda_time": 0.045847701115467716, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 373.698, - "cuda_time_us": 781.887, - "pct_cuda_time": 0.8213100896297503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.169, - "cuda_time_us": 270.65599999999995, - "pct_cuda_time": 0.28430259566769833, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.888, - "pct_cuda_time": 0.28349587276677324, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 115.651, - "cuda_time_us": 54.944, - "pct_cuda_time": 0.05771430087035211, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.944, - "pct_cuda_time": 0.05771430087035211, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.627, - "cuda_time_us": 264.79900000000004, - "pct_cuda_time": 0.2781502831277004, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.463, - "pct_cuda_time": 0.02359559443161618, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.671, - "pct_cuda_time": 0.25280573865696915, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.665, - "pct_cuda_time": 0.001748950039115031, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.103, - "cuda_time_us": 191.488, - "pct_cuda_time": 0.20114290996399942, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011428574429772696, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 190.4, - "pct_cuda_time": 0.20000005252102218, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.515, - "cuda_time_us": 44.448, - "pct_cuda_time": 0.04668908789104198, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.448, - "pct_cuda_time": 0.04668908789104198, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.183, - "cuda_time_us": 2022.6509999999998, - "pct_cuda_time": 2.1246339613009346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.105, - "cuda_time_us": 1267.3890000000001, - "pct_cuda_time": 1.3312913159903663, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1266.653, - "pct_cuda_time": 1.3305182065436465, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.789, - "cuda_time_us": 174.303, - "pct_cuda_time": 0.18309143463535568, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.303, - "pct_cuda_time": 0.18309143463535568, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.916, - "cuda_time_us": 580.959, - "pct_cuda_time": 0.6102512106752127, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 580.223, - "pct_cuda_time": 0.6094781012284928, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 497.529, - "cuda_time_us": 2890.934, - "pct_cuda_time": 3.036696175602987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.175, - "cuda_time_us": 43.936, - "pct_cuda_time": 0.04615127262375855, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.936, - "pct_cuda_time": 0.04615127262375855, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 357.207, - "cuda_time_us": 780.988, - "pct_cuda_time": 0.8203657616506725, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.953, - "cuda_time_us": 272.063, - "pct_cuda_time": 0.28578053723228386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0010084036261564142, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 271.103, - "pct_cuda_time": 0.28477213360612746, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.11, - "cuda_time_us": 53.888, - "pct_cuda_time": 0.056605056881580046, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.888, - "pct_cuda_time": 0.056605056881580046, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 148.878, - "cuda_time_us": 265.183, - "pct_cuda_time": 0.2785536445781629, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.304, - "pct_cuda_time": 0.023428577581034022, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 241.567, - "pct_cuda_time": 0.25374691537471517, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.0013781516224137661, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.415, - "cuda_time_us": 189.854, - "pct_cuda_time": 0.19942652295864569, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007720590262760047, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.119, - "pct_cuda_time": 0.19865446393236966, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.619, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.04611765916955335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.04611765916955335, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.559, - "cuda_time_us": 2022.106, - "pct_cuda_time": 2.1240614821590023, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.44, - "cuda_time_us": 1265.117, - "pct_cuda_time": 1.328904760741796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011092439887720558, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1264.061, - "pct_cuda_time": 1.327795516753024, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.929, - "cuda_time_us": 173.952, - "pct_cuda_time": 0.18272273705954226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 173.952, - "pct_cuda_time": 0.18272273705954226, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.743, - "cuda_time_us": 583.037, - "pct_cuda_time": 0.6124339843576638, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007720590262760047, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.302, - "pct_cuda_time": 0.6116619253313879, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 510.321, - "cuda_time_us": 2893.3679999999995, - "pct_cuda_time": 3.03925289896347, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.627, - "cuda_time_us": 44.832, - "pct_cuda_time": 0.04709244934150454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.832, - "pct_cuda_time": 0.04709244934150454, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 361.755, - "cuda_time_us": 781.0539999999999, - "pct_cuda_time": 0.8204350893999707, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.153, - "cuda_time_us": 272.352, - "pct_cuda_time": 0.2860841087405747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.0014453785308241936, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 270.976, - "pct_cuda_time": 0.2846387302097505, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.032, - "cuda_time_us": 54.017, - "pct_cuda_time": 0.056740561118844825, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.017, - "pct_cuda_time": 0.056740561118844825, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.219, - "cuda_time_us": 264.702, - "pct_cuda_time": 0.2780483923446408, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.623, - "pct_cuda_time": 0.02376366170264225, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.736, - "pct_cuda_time": 0.25287401598582343, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.001410714656175067, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.415, - "cuda_time_us": 189.983, - "pct_cuda_time": 0.19956202719591046, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.216, - "pct_cuda_time": 0.0012773112597981246, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.767, - "pct_cuda_time": 0.19828471593611235, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.722, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.046016818806937705, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.046016818806937705, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.802, - "cuda_time_us": 2023.6739999999998, - "pct_cuda_time": 2.125708541415057, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.684, - "cuda_time_us": 1264.6999999999998, - "pct_cuda_time": 1.3284667354166841, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007720590262760047, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1263.965, - "pct_cuda_time": 1.3276946763904083, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.156, - "cuda_time_us": 174.271, - "pct_cuda_time": 0.18305782118115047, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.271, - "pct_cuda_time": 0.18305782118115047, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.317, - "cuda_time_us": 584.703, - "pct_cuda_time": 0.6141839848172227, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011428574429772696, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 583.615, - "pct_cuda_time": 0.6130411273742455, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 506.453, - "cuda_time_us": 2886.967, - "pct_cuda_time": 3.032529157701984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.579, - "cuda_time_us": 44.672, - "pct_cuda_time": 0.04692438207047847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.672, - "pct_cuda_time": 0.04692438207047847, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 361.843, - "cuda_time_us": 777.758, - "pct_cuda_time": 0.8169729036168338, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.09, - "cuda_time_us": 269.471, - "pct_cuda_time": 0.28305784744166157, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 268.735, - "pct_cuda_time": 0.28228473799494164, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.716, - "cuda_time_us": 53.92, - "pct_cuda_time": 0.05663867033578526, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.92, - "pct_cuda_time": 0.05663867033578526, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.414, - "cuda_time_us": 264.767, - "pct_cuda_time": 0.27811666967349513, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.592, - "pct_cuda_time": 0.023731098668880946, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.671, - "pct_cuda_time": 0.25280573865696915, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.001579832347645049, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.245, - "cuda_time_us": 189.6, - "pct_cuda_time": 0.1991597161658918, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.0014453785308241936, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.224, - "pct_cuda_time": 0.1977143376350676, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.154, - "cuda_time_us": 43.103, - "pct_cuda_time": 0.045276272393979085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.103, - "pct_cuda_time": 0.045276272393979085, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.526, - "cuda_time_us": 2021.4340000000002, - "pct_cuda_time": 2.123355599620693, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.458, - "cuda_time_us": 1264.604, - "pct_cuda_time": 1.3283658950540689, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.055, - "pct_cuda_time": 0.0011081935683281427, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1263.549, - "pct_cuda_time": 1.3272577014857405, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.975, - "cuda_time_us": 174.079, - "pct_cuda_time": 0.18285614045591922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.079, - "pct_cuda_time": 0.18285614045591922, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.949, - "cuda_time_us": 582.7510000000001, - "pct_cuda_time": 0.6121335641107049, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.00141176507661898, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.407, - "pct_cuda_time": 0.6107217990340857, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 481.086, - "cuda_time_us": 2891.4800000000005, - "pct_cuda_time": 3.0372697051653637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.318, - "cuda_time_us": 44.544, - "pct_cuda_time": 0.04678992825365762, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.544, - "pct_cuda_time": 0.04678992825365762, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 339.956, - "cuda_time_us": 779.389, - "pct_cuda_time": 0.8186861393608558, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.614, - "cuda_time_us": 270.207, - "pct_cuda_time": 0.2838309568883815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 269.471, - "pct_cuda_time": 0.28305784744166157, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.132, - "cuda_time_us": 54.24, - "pct_cuda_time": 0.056974804877837404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.24, - "pct_cuda_time": 0.056974804877837404, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 147.014, - "cuda_time_us": 265.40700000000004, - "pct_cuda_time": 0.2787889387575994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.656, - "pct_cuda_time": 0.023798325577291373, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 241.055, - "pct_cuda_time": 0.2532091001074317, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.696, - "pct_cuda_time": 0.0017815130728763317, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.151, - "cuda_time_us": 189.535, - "pct_cuda_time": 0.19909143883703748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011428574429772696, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 188.447, - "pct_cuda_time": 0.19794858139406019, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.522, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.045680684264885564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.045680684264885564, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.426, - "cuda_time_us": 2024.0590000000002, - "pct_cuda_time": 2.1261129532859644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.173, - "cuda_time_us": 1265.853, - "pct_cuda_time": 1.3296778701885161, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1265.117, - "pct_cuda_time": 1.328904760741796, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.207, - "cuda_time_us": 175.104, - "pct_cuda_time": 0.18393282141092995, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 175.104, - "pct_cuda_time": 0.18393282141092995, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.376, - "cuda_time_us": 583.102, - "pct_cuda_time": 0.6125022616865181, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011764708971824835, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 581.982, - "pct_cuda_time": 0.6113257907893357, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 530.572, - "cuda_time_us": 2893.7209999999995, - "pct_cuda_time": 3.0396236973801716, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.32, - "cuda_time_us": 43.583, - "pct_cuda_time": 0.04578047420705729, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.583, - "pct_cuda_time": 0.04578047420705729, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 380.175, - "cuda_time_us": 780.6379999999999, - "pct_cuda_time": 0.8199981144953029, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.39, - "cuda_time_us": 271.551, - "pct_cuda_time": 0.28524272196500045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007720590262760047, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 270.816, - "pct_cuda_time": 0.2844706629387244, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.855, - "cuda_time_us": 54.048, - "pct_cuda_time": 0.05677312415260612, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.048, - "pct_cuda_time": 0.05677312415260612, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 166.59, - "cuda_time_us": 264.799, - "pct_cuda_time": 0.27815028312770035, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.464, - "pct_cuda_time": 0.023596644852060093, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.863, - "pct_cuda_time": 0.25300741938220045, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.0015462188934398352, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.711, - "cuda_time_us": 190.23999999999998, - "pct_cuda_time": 0.19983198524999604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.504, - "pct_cuda_time": 0.19905887580327616, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.503, - "cuda_time_us": 43.552, - "pct_cuda_time": 0.04574791117329599, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.552, - "pct_cuda_time": 0.04574791117329599, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.997, - "cuda_time_us": 2025.9479999999999, - "pct_cuda_time": 2.1280971975045153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.484, - "cuda_time_us": 1268.062, - "pct_cuda_time": 1.3319982489491198, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0014789919850294075, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1266.654, - "pct_cuda_time": 1.3305192569640902, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.161, - "cuda_time_us": 174.783, - "pct_cuda_time": 0.18359563644843388, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.783, - "pct_cuda_time": 0.18359563644843388, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.055, - "cuda_time_us": 583.103, - "pct_cuda_time": 0.6125033121069621, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.367, - "pct_cuda_time": 0.6117302026602421, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 504.368, - "cuda_time_us": 2888.059, - "pct_cuda_time": 3.0336762168267373, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.053, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.04578152462750121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.04578152462750121, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 360.172, - "cuda_time_us": 778.495, - "pct_cuda_time": 0.8177470634839976, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.913, - "cuda_time_us": 269.887, - "pct_cuda_time": 0.28349482234632933, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 268.863, - "pct_cuda_time": 0.2824191918117625, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.704, - "cuda_time_us": 54.144, - "pct_cuda_time": 0.05687396451522177, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.144, - "pct_cuda_time": 0.05687396451522177, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.742, - "cuda_time_us": 264.352, - "pct_cuda_time": 0.27768074518927127, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.02403361975672787, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.128, - "pct_cuda_time": 0.2522353603559244, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.00141176507661898, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.174, - "cuda_time_us": 190.112, - "pct_cuda_time": 0.19969753143317523, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0011102944092159686, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.055, - "pct_cuda_time": 0.19858723702395925, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.077, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.04547900353965428, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.04547900353965428, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.604, - "cuda_time_us": 2022.684, - "pct_cuda_time": 2.124668625175584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.16, - "cuda_time_us": 1264.221, - "pct_cuda_time": 1.3279635840240502, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1263.485, - "pct_cuda_time": 1.3271904745773302, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.809, - "cuda_time_us": 174.847, - "pct_cuda_time": 0.18366286335684434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.847, - "pct_cuda_time": 0.18366286335684434, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.664, - "cuda_time_us": 583.616, - "pct_cuda_time": 0.6130421777946894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.185, - "pct_cuda_time": 0.001244748226036824, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 582.431, - "pct_cuda_time": 0.6117974295686527, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 487.127, - "cuda_time_us": 2892.026, - "pct_cuda_time": 3.0378432347277395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.862, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.04611765916955335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.04611765916955335, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 350.574, - "cuda_time_us": 780.0300000000001, - "pct_cuda_time": 0.8193594588654041, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.098, - "cuda_time_us": 270.784, - "pct_cuda_time": 0.28443704948451926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 270.048, - "pct_cuda_time": 0.2836639400377993, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.322, - "cuda_time_us": 53.92, - "pct_cuda_time": 0.05663867033578526, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.92, - "pct_cuda_time": 0.05663867033578526, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.635, - "cuda_time_us": 265.024, - "pct_cuda_time": 0.27838662772758077, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.816, - "pct_cuda_time": 0.023966392848317444, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.512, - "pct_cuda_time": 0.25263872180638697, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.696, - "pct_cuda_time": 0.0017815130728763317, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.524, - "cuda_time_us": 190.30200000000002, - "pct_cuda_time": 0.19989711131751872, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007720590262760047, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.567, - "pct_cuda_time": 0.1991250522912427, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.869, - "cuda_time_us": 42.688, - "pct_cuda_time": 0.04484034790975522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 42.688, - "pct_cuda_time": 0.04484034790975522, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 92.808, - "cuda_time_us": 2025.404, - "pct_cuda_time": 2.127525768783027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.394, - "cuda_time_us": 1265.981, - "pct_cuda_time": 1.329812324005337, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011428574429772696, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1264.893, - "pct_cuda_time": 1.3286694665623595, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.215, - "cuda_time_us": 174.848, - "pct_cuda_time": 0.18366391377728827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 174.848, - "pct_cuda_time": 0.18366391377728827, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.212, - "cuda_time_us": 584.5749999999999, - "pct_cuda_time": 0.6140495310004018, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.0014453785308241936, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 583.199, - "pct_cuda_time": 0.6126041524695777, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 494.076, - "cuda_time_us": 3203.833, - "pct_cuda_time": 3.3653716820828987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.793, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.046016818806937705, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.046016818806937705, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 349.317, - "cuda_time_us": 781.087, - "pct_cuda_time": 0.8204697532746198, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.899, - "cuda_time_us": 271.392, - "pct_cuda_time": 0.28507570511441827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011092439887720558, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 270.336, - "pct_cuda_time": 0.2839664611256463, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 96.661, - "cuda_time_us": 53.952, - "pct_cuda_time": 0.05667228378999048, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.952, - "pct_cuda_time": 0.05667228378999048, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 150.931, - "cuda_time_us": 264.863, - "pct_cuda_time": 0.2782175100361108, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.463, - "pct_cuda_time": 0.02359559443161618, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 240.703, - "pct_cuda_time": 0.25283935211117436, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.697, - "pct_cuda_time": 0.0017825634933202447, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.047, - "cuda_time_us": 190.88, - "pct_cuda_time": 0.20050425433410035, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 190.112, - "pct_cuda_time": 0.19969753143317523, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.721, - "cuda_time_us": 44.64, - "pct_cuda_time": 0.04689076861627326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.64, - "pct_cuda_time": 0.04689076861627326, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.133, - "cuda_time_us": 2334.2980000000002, - "pct_cuda_time": 2.4519943413850687, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.832, - "cuda_time_us": 1479.6760000000002, - "pct_cuda_time": 1.5542819207673109, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1478.94, - "pct_cuda_time": 1.5535088113205908, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.526, - "cuda_time_us": 186.144, - "pct_cuda_time": 0.19552946311172872, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 186.144, - "pct_cuda_time": 0.19552946311172872, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.294, - "cuda_time_us": 668.4780000000001, - "pct_cuda_time": 0.7021829575060287, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011092439887720558, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 667.422, - "pct_cuda_time": 0.7010737135172566, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 501.846, - "cuda_time_us": 3342.331, - "pct_cuda_time": 3.5108528127239524, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.015, - "cuda_time_us": 44.672, - "pct_cuda_time": 0.04692438207047847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.672, - "pct_cuda_time": 0.04692438207047847, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 356.278, - "cuda_time_us": 911.0390000000001, - "pct_cuda_time": 0.9569739908019933, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.694, - "cuda_time_us": 312.0, - "pct_cuda_time": 0.3277311785008346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 311.264, - "pct_cuda_time": 0.32695806905411473, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.252, - "cuda_time_us": 58.944, - "pct_cuda_time": 0.06191598264600383, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 58.944, - "pct_cuda_time": 0.06191598264600383, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.141, - "cuda_time_us": 317.024, - "pct_cuda_time": 0.33300849081105316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 24.064, - "pct_cuda_time": 0.025277317562320784, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 291.488, - "pct_cuda_time": 0.3061849543552925, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.0015462188934398352, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.667, - "cuda_time_us": 223.071, - "pct_cuda_time": 0.23431833884410155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 222.335, - "pct_cuda_time": 0.23354522939738162, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.255, - "cuda_time_us": 44.704, - "pct_cuda_time": 0.04695799552468369, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.704, - "pct_cuda_time": 0.04695799552468369, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.368, - "cuda_time_us": 2341.916, - "pct_cuda_time": 2.459996444326797, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.445, - "cuda_time_us": 1487.741, - "pct_cuda_time": 1.5627535616474686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.001075630534566842, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1486.717, - "pct_cuda_time": 1.5616779311129019, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.028, - "cuda_time_us": 186.272, - "pct_cuda_time": 0.1956639169285496, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 186.272, - "pct_cuda_time": 0.1956639169285496, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.9, - "cuda_time_us": 667.903, - "pct_cuda_time": 0.7015789657507787, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 667.167, - "pct_cuda_time": 0.7008058563040588, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 499.387, - "cuda_time_us": 3345.6899999999996, - "pct_cuda_time": 3.514381174995055, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.544, - "cuda_time_us": 45.343, - "pct_cuda_time": 0.04762921418834406, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.343, - "pct_cuda_time": 0.04762921418834406, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 348.963, - "cuda_time_us": 913.343, - "pct_cuda_time": 0.9593941595047685, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.813, - "cuda_time_us": 313.472, - "pct_cuda_time": 0.3292773973942744, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015126054392346213, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 312.032, - "pct_cuda_time": 0.3277647919550398, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 96.212, - "cuda_time_us": 58.975, - "pct_cuda_time": 0.06194854567976514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 58.975, - "pct_cuda_time": 0.06194854567976514, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 150.812, - "cuda_time_us": 317.217, - "pct_cuda_time": 0.33321122195672837, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 24.321, - "pct_cuda_time": 0.025547275616406406, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 291.424, - "pct_cuda_time": 0.30611772744688215, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.0015462188934398352, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.116, - "cuda_time_us": 223.679, - "pct_cuda_time": 0.2349569944740006, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011092439887720558, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 222.623, - "pct_cuda_time": 0.23384775048522852, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.749, - "cuda_time_us": 45.92, - "pct_cuda_time": 0.04823530678448182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.92, - "pct_cuda_time": 0.04823530678448182, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.536, - "cuda_time_us": 2341.084, - "pct_cuda_time": 2.459122494517461, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.958, - "cuda_time_us": 1487.133, - "pct_cuda_time": 1.5621149060175694, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1486.365, - "pct_cuda_time": 1.5613081831166444, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.749, - "cuda_time_us": 186.527, - "pct_cuda_time": 0.19593177414174734, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 186.527, - "pct_cuda_time": 0.19593177414174734, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.092, - "cuda_time_us": 667.424, - "pct_cuda_time": 0.7010758143581444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.025, - "pct_cuda_time": 0.0010766809550107547, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 666.399, - "pct_cuda_time": 0.6999991334031336, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 505.974, - "cuda_time_us": 3338.263, - "pct_cuda_time": 3.506579702358114, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.638, - "cuda_time_us": 45.984, - "pct_cuda_time": 0.04830253369289224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.984, - "pct_cuda_time": 0.04830253369289224, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 362.856, - "cuda_time_us": 909.6619999999999, - "pct_cuda_time": 0.955527561850725, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.394, - "cuda_time_us": 310.687, - "pct_cuda_time": 0.326351976457977, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 309.951, - "pct_cuda_time": 0.32557886701125704, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.38, - "cuda_time_us": 58.528, - "pct_cuda_time": 0.06147900774133605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 58.528, - "pct_cuda_time": 0.06147900774133605, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.531, - "cuda_time_us": 316.671, - "pct_cuda_time": 0.3326376923943519, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 24.256, - "pct_cuda_time": 0.025478998287552068, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 290.783, - "pct_cuda_time": 0.30544440794233396, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.001714286164465904, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[2], int32[2], None, None, None, 4096, 4096, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.344, - "cuda_time_us": 223.77599999999998, - "pct_cuda_time": 0.23505888525706012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0014789919850294075, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 222.368, - "pct_cuda_time": 0.23357989327203071, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.488, - "cuda_time_us": 44.256, - "pct_cuda_time": 0.04648740716581069, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.256, - "pct_cuda_time": 0.04648740716581069, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.544, - "cuda_time_us": 2338.361, - "pct_cuda_time": 2.456262199648686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.686, - "cuda_time_us": 1486.971, - "pct_cuda_time": 1.5619447379056557, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.087, - "pct_cuda_time": 0.0011418070225333564, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1485.884, - "pct_cuda_time": 1.5608029308831224, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.706, - "cuda_time_us": 186.047, - "pct_cuda_time": 0.19542757232866914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 186.047, - "pct_cuda_time": 0.19542757232866914, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.501, - "cuda_time_us": 665.343, - "pct_cuda_time": 0.6988898894143616, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007731094467199176, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 664.607, - "pct_cuda_time": 0.6981167799676417, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.776, - "cuda_time_us": 45.407, - "pct_cuda_time": 0.04769644109675447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.407, - "pct_cuda_time": 0.04769644109675447, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 91.588, - "cuda_time_us": 356.608, - "pct_cuda_time": 0.3745883336629027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.0034296227493757214, - "trace": "index_select(bfloat16[4096, 4096], 0, int64[1])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008067229009251313, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[1, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 352.575, - "pct_cuda_time": 0.37035198801260183, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[1, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 78884.866, - "cuda_time_us": 144.192, - "pct_cuda_time": 0.1514622246486934, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.00352941269154745, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.0022857148859545392, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.0022857148859545392, - "trace": "copy_(int32[1], int32[1], True) <- _to_copy(int32[1], 3, 0, None, None, True, None) <- to(int32[1], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.0023193283401597526, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.0023193283401597526, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.0022857148859545392, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.002352941794364967, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.704, - "pct_cuda_time": 0.0049411777681664295, - "trace": "copy_(float32[1, 128256], bfloat16[1, 128256], False) <- _to_copy(bfloat16[1, 128256], 6, None, None, None, False, None) <- to(bfloat16[1, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 5.376, - "pct_cuda_time": 0.00564706030647592, - "trace": "div_(float32[1, 128256], bfloat16[1, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.64, - "pct_cuda_time": 0.04268908684062153, - "trace": "_softmax(float32[1, 128256], -1, False) <- softmax(float32[1, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.576, - "pct_cuda_time": 0.034218496380907654, - "trace": "_log_softmax(float32[1, 128256], -1, False) <- log_softmax(float32[1, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.0022521014317493254, - "trace": "copy_(int64[1], int32[1], False) <- _to_copy(int32[1], 4, None, None, None, False, None) <- to(int32[1], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 5.312, - "pct_cuda_time": 0.005579833398065492, - "trace": "index(float32[1, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 33.728, - "pct_cuda_time": 0.03542858073229535, - "trace": "argmax(float32[1, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.003327731966316167, - "trace": "copy_(int64[1], int64[1], False) <- _to_copy(int64[1], 4, 0, None, None, False, None) <- to(int64[1], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - }, - "decode_1": { - "metadata": { - "num_running_seqs": 1 - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 6791.93, - "pct_cuda_time": 93.5046501558771, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 3.104, - "pct_cuda_time": 0.0427328364815071, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.104, - "pct_cuda_time": 0.0427328364815071, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 6785.37, - "pct_cuda_time": 93.41433849114814, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 222.2410000000001, - "pct_cuda_time": 3.059596750156773, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.576, - "pct_cuda_time": 0.06299789295727334, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 217.6650000000001, - "pct_cuda_time": 2.9965988571995, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 2328.683, - "pct_cuda_time": 32.059030237198904, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 664.3159999999997, - "pct_cuda_time": 9.145653028366256, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 664.3159999999997, - "pct_cuda_time": 9.145653028366256, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 119.84, - "pct_cuda_time": 1.64983992395097, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 119.84, - "pct_cuda_time": 1.64983992395097, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 956.726, - "pct_cuda_time": 13.171267949615451, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 74.11200000000001, - "pct_cuda_time": 1.0203015390842314, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cuda_time_us": 753.1120000000001, - "pct_cuda_time": 10.368109519413911, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cuda_time_us": 81.40700000000001, - "pct_cuda_time": 1.1207319650290106, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 48.095, - "pct_cuda_time": 0.6621249260883003, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 587.8009999999999, - "pct_cuda_time": 8.092269335266222, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cuda_time_us": 587.8009999999999, - "pct_cuda_time": 8.092269335266222, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 4234.446, - "pct_cuda_time": 58.29571150379247, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 2560.629, - "pct_cuda_time": 35.2522359364707, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2560.629, - "pct_cuda_time": 35.2522359364707, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 282.43, - "pct_cuda_time": 3.8882200410670267, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 282.43, - "pct_cuda_time": 3.8882200410670267, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 1391.3869999999997, - "pct_cuda_time": 19.155255526254738, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1391.3869999999997, - "pct_cuda_time": 19.155255526254738, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 347.518, - "pct_cuda_time": 4.784287973060691, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 2.624, - "pct_cuda_time": 0.03612466589158332, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.735, - "pct_cuda_time": 0.010118761215820785, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 344.159, - "pct_cuda_time": 4.7380445459532865, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 124.28700000000002, - "pct_cuda_time": 1.7110618710622016, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 16.031, - "pct_cuda_time": 0.22069913068139183, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.192, - "pct_cuda_time": 0.057711356485334334, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.512, - "pct_cuda_time": 0.062116803545283504, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 34.08, - "pct_cuda_time": 0.46918011188458825, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.584, - "pct_cuda_time": 0.3797495365676198, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 1.888, - "pct_cuda_time": 0.025992137653700193, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.64, - "pct_cuda_time": 0.06387898236926319, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 28.576, - "pct_cuda_time": 0.3934064224534623, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.784, - "pct_cuda_time": 0.03832738942155792, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 17089.731, - "cuda_time_us": 6791.93, - "pct_cuda_time": 93.5046501558771, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 68.255, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.0427328364815071, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.0427328364815071, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[1]) <- embedding(bfloat16[128256, 4096], int64[1], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 833.35, - "cuda_time_us": 214.909, - "pct_cuda_time": 2.9586569443956856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 50.825, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.06299789295727334, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.06299789295727334, - "trace": "_C::rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 601.687, - "cuda_time_us": 74.464, - "pct_cuda_time": 1.0251475308501754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 93.019, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.31234619655039725, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.31234619655039725, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 182.562, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 223.755, - "cuda_time_us": 29.599999999999998, - "pct_cuda_time": 0.4075038530452996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.048, - "pct_cuda_time": 0.02819486118367479, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.32468144831825496, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921942361608726, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 46.424, - "cuda_time_us": 18.4, - "pct_cuda_time": 0.2533132059470782, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.4, - "pct_cuda_time": 0.2533132059470782, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.425, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04537610471747661, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04537610471747661, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 128.22, - "cuda_time_us": 132.573, - "pct_cuda_time": 1.8251354158707607, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 46.138, - "cuda_time_us": 79.871, - "pct_cuda_time": 1.0995858191412542, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.871, - "pct_cuda_time": 1.0995858191412542, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 29.968, - "cuda_time_us": 8.831, - "pct_cuda_time": 0.12157657183253519, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.831, - "pct_cuda_time": 0.12157657183253519, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 34.706, - "cuda_time_us": 43.871, - "pct_cuda_time": 0.603973024896971, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.871, - "pct_cuda_time": 0.603973024896971, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 547.732, - "cuda_time_us": 214.399, - "pct_cuda_time": 2.9516357631438916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.372, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.050662641189415644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.050662641189415644, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 391.377, - "cuda_time_us": 73.08699999999999, - "pct_cuda_time": 1.0061903414703315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.464, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.2850324247787123, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.2850324247787123, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 118.811, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.052865364719390226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.052865364719390226, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 176.711, - "cuda_time_us": 30.174999999999997, - "pct_cuda_time": 0.4154198907311458, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03215976353762905, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.807, - "pct_cuda_time": 0.327751494238157, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03524357647959349, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.020265056475766253, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 32.192, - "cuda_time_us": 18.368, - "pct_cuda_time": 0.2528726612410832, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.368, - "pct_cuda_time": 0.2528726612410832, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.703, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.285, - "cuda_time_us": 134.208, - "pct_cuda_time": 1.8476444969426884, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.2, - "cuda_time_us": 82.208, - "pct_cuda_time": 1.1317593497009457, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.208, - "pct_cuda_time": 1.1317593497009457, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.84, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12159033885459752, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12159033885459752, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.221, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.5942948083871451, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.5942948083871451, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 504.409, - "cuda_time_us": 215.231, - "pct_cuda_time": 2.96308992549976, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.327, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 357.289, - "cuda_time_us": 72.28699999999999, - "pct_cuda_time": 0.9951767238204586, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.483, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.2872351483086869, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.2872351483086869, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 107.246, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.050222096483420714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.050222096483420714, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.445, - "cuda_time_us": 29.535, - "pct_cuda_time": 0.4066089966112475, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.031278674125639214, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.327, - "pct_cuda_time": 0.3211433236482333, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.465, - "pct_cuda_time": 0.03393570938367107, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.471, - "pct_cuda_time": 0.020251289453703913, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.529, - "cuda_time_us": 18.24, - "pct_cuda_time": 0.25111048241710354, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.24, - "pct_cuda_time": 0.25111048241710354, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.988, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.376, - "cuda_time_us": 136.12800000000001, - "pct_cuda_time": 1.8740771793023838, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.99, - "cuda_time_us": 83.488, - "pct_cuda_time": 1.1493811379407426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 83.488, - "pct_cuda_time": 1.1493811379407426, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.968, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12026870473661277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12026870473661277, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.517, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.6044273366250283, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.6044273366250283, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 502.608, - "cuda_time_us": 212.223, - "pct_cuda_time": 2.921678723136238, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.855, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 355.405, - "cuda_time_us": 72.864, - "pct_cuda_time": 1.0031202955504297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.179, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.2832702459547326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.2832702459547326, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.954, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.704, - "cuda_time_us": 29.951999999999998, - "pct_cuda_time": 0.41234984481124376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.32468144831825496, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.035684121185588405, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.021146145887756092, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.51, - "cuda_time_us": 18.56, - "pct_cuda_time": 0.25551592947705276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.56, - "pct_cuda_time": 0.25551592947705276, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.002, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.234, - "cuda_time_us": 132.607, - "pct_cuda_time": 1.8256034946208803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.115, - "cuda_time_us": 80.223, - "pct_cuda_time": 1.1044318109071984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.223, - "pct_cuda_time": 1.1044318109071984, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.587, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.11806598120663817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.11806598120663817, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.383, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6031057025070434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6031057025070434, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 492.75, - "cuda_time_us": 211.551, - "pct_cuda_time": 2.9124272843103443, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.707, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.048900462365435965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.048900462365435965, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 349.782, - "cuda_time_us": 72.576, - "pct_cuda_time": 0.9991553931964752, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.905, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.999, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.050662641189415644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.050662641189415644, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.034, - "cuda_time_us": 30.08, - "pct_cuda_time": 0.4141120236352235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.033040852949618886, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.712, - "pct_cuda_time": 0.3264436271422346, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921942361608726, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 26.756, - "cuda_time_us": 18.176, - "pct_cuda_time": 0.2502293930051137, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.176, - "pct_cuda_time": 0.2502293930051137, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.918, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04537610471747661, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04537610471747661, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.955, - "cuda_time_us": 132.12699999999998, - "pct_cuda_time": 1.8189953240309562, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.156, - "cuda_time_us": 80.064, - "pct_cuda_time": 1.1022428543992862, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.064, - "pct_cuda_time": 1.1022428543992862, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.186, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.11806598120663817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.11806598120663817, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.567, - "cuda_time_us": 43.487, - "pct_cuda_time": 0.598686488425032, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.487, - "pct_cuda_time": 0.598686488425032, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 535.017, - "cuda_time_us": 211.965, - "pct_cuda_time": 2.9181268314441535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 20.186, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 373.628, - "cuda_time_us": 72.288, - "pct_cuda_time": 0.995190490842521, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.615, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.361, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 175.325, - "cuda_time_us": 29.568, - "pct_cuda_time": 0.40706330833930476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.296, - "pct_cuda_time": 0.3207165459643007, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.034803031773598565, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 30.109, - "cuda_time_us": 18.304, - "pct_cuda_time": 0.2519915718290934, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.304, - "pct_cuda_time": 0.2519915718290934, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.501, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.116, - "cuda_time_us": 132.733, - "pct_cuda_time": 1.8273381394007353, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.3, - "cuda_time_us": 80.447, - "pct_cuda_time": 1.107515623849163, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.447, - "pct_cuda_time": 1.107515623849163, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.014, - "cuda_time_us": 9.151, - "pct_cuda_time": 0.12598201889248437, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.151, - "pct_cuda_time": 0.12598201889248437, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.411, - "cuda_time_us": 43.135, - "pct_cuda_time": 0.5938404966590878, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.135, - "pct_cuda_time": 0.5938404966590878, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 526.821, - "cuda_time_us": 211.774, - "pct_cuda_time": 2.9154973302302465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.021, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 368.943, - "cuda_time_us": 72.606, - "pct_cuda_time": 0.9995684038583456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.847, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.2806269777187631, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.2806269777187631, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.361, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.051103185895410554, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.051103185895410554, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 169.301, - "cuda_time_us": 29.854, - "pct_cuda_time": 0.4110006766491343, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.271, - "pct_cuda_time": 0.03126490710357688, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.32468144831825496, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436248706760365, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.02069183415969883, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 28.865, - "cuda_time_us": 18.656, - "pct_cuda_time": 0.2568375635950375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.656, - "pct_cuda_time": 0.2568375635950375, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.283, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04537610471747661, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04537610471747661, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 106.243, - "cuda_time_us": 132.38400000000001, - "pct_cuda_time": 1.8225334487009783, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.521, - "cuda_time_us": 80.608, - "pct_cuda_time": 1.1097321144011998, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.608, - "pct_cuda_time": 1.1097321144011998, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 27.938, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.144, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.5938542636811502, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.5938542636811502, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 515.051, - "cuda_time_us": 211.99800000000002, - "pct_cuda_time": 2.9185811431722106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.298, - "cuda_time_us": 3.873, - "pct_cuda_time": 0.05331967644744749, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.873, - "pct_cuda_time": 0.05331967644744749, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 367.597, - "cuda_time_us": 72.31700000000001, - "pct_cuda_time": 0.995589734482329, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.727, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 112.782, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05154373060140547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05154373060140547, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 167.782, - "cuda_time_us": 29.598000000000003, - "pct_cuda_time": 0.40747631900117504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.359, - "pct_cuda_time": 0.3215838683542282, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436248706760365, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.02069183415969883, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.076, - "cuda_time_us": 18.335, - "pct_cuda_time": 0.252418349513026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.335, - "pct_cuda_time": 0.252418349513026, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.261, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.96, - "cuda_time_us": 132.352, - "pct_cuda_time": 1.8220929039949831, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.04, - "cuda_time_us": 79.488, - "pct_cuda_time": 1.0943130496913775, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.488, - "pct_cuda_time": 1.0943130496913775, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.635, - "cuda_time_us": 8.48, - "pct_cuda_time": 0.11674434708865342, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.48, - "pct_cuda_time": 0.11674434708865342, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.843, - "cuda_time_us": 44.384, - "pct_cuda_time": 0.611035507214952, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.384, - "pct_cuda_time": 0.611035507214952, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 492.119, - "cuda_time_us": 211.968, - "pct_cuda_time": 2.91816813251034, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.119, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.049341007071430874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.049341007071430874, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 337.822, - "cuda_time_us": 72.097, - "pct_cuda_time": 0.9925609896286137, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.866, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.281508067130753, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.281508067130753, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.297, - "cuda_time_us": 3.585, - "pct_cuda_time": 0.049354774093493224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.585, - "pct_cuda_time": 0.049354774093493224, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.941, - "cuda_time_us": 29.951999999999998, - "pct_cuda_time": 0.41234984481124376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171921883163414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.552, - "pct_cuda_time": 0.32424090361226005, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.035684121185588405, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 25.766, - "cuda_time_us": 18.112, - "pct_cuda_time": 0.24934830359312388, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.112, - "pct_cuda_time": 0.24934830359312388, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.425, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 105.429, - "cuda_time_us": 132.927, - "pct_cuda_time": 1.8300089416808292, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 39.223, - "cuda_time_us": 80.223, - "pct_cuda_time": 1.1044318109071984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.223, - "pct_cuda_time": 1.1044318109071984, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.616, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.197, - "cuda_time_us": 44.064, - "pct_cuda_time": 0.6066300601550029, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.064, - "pct_cuda_time": 0.6066300601550029, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 515.681, - "cuda_time_us": 211.423, - "pct_cuda_time": 2.9106651054863644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.273, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.048900462365435965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.048900462365435965, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 370.649, - "cuda_time_us": 73.055, - "pct_cuda_time": 1.005749796764337, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 47.811, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.2876756930146818, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.2876756930146818, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.45, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242482001339531, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242482001339531, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.927, - "cuda_time_us": 29.950999999999997, - "pct_cuda_time": 0.4123360777891814, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03215976353762905, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.32468144831825496, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.034803031773598565, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.02069183415969883, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 28.958, - "cuda_time_us": 18.4, - "pct_cuda_time": 0.2533132059470782, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.4, - "pct_cuda_time": 0.2533132059470782, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.336, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.163, - "cuda_time_us": 131.392, - "pct_cuda_time": 1.8088765628151358, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.053, - "cuda_time_us": 79.168, - "pct_cuda_time": 1.0899076026314287, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.168, - "pct_cuda_time": 1.0899076026314287, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.191, - "cuda_time_us": 9.184, - "pct_cuda_time": 0.1264363306205416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.184, - "pct_cuda_time": 0.1264363306205416, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.545, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.5925326295631654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.5925326295631654, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 488.661, - "cuda_time_us": 211.67700000000002, - "pct_cuda_time": 2.9141619290901994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.692, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 341.136, - "cuda_time_us": 72.381, - "pct_cuda_time": 0.9964708238943187, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.736, - "cuda_time_us": 20.671, - "pct_cuda_time": 0.28457811305065506, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.671, - "pct_cuda_time": 0.28457811305065506, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.335, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.050222096483420714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.050222096483420714, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.053, - "cuda_time_us": 29.854999999999997, - "pct_cuda_time": 0.41101444367119666, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03215976353762905, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.423, - "pct_cuda_time": 0.32246495776621803, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.035684121185588405, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.12, - "cuda_time_us": 18.207, - "pct_cuda_time": 0.25065617068904633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.207, - "pct_cuda_time": 0.25065617068904633, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.955, - "cuda_time_us": 3.457, - "pct_cuda_time": 0.047592595269513545, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.457, - "pct_cuda_time": 0.047592595269513545, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.799, - "cuda_time_us": 132.223, - "pct_cuda_time": 1.8203169581489413, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.565, - "cuda_time_us": 79.103, - "pct_cuda_time": 1.0890127461973762, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.103, - "pct_cuda_time": 1.0890127461973762, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.375, - "cuda_time_us": 9.536, - "pct_cuda_time": 0.13128232238648574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.536, - "pct_cuda_time": 0.13128232238648574, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.177, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.6000218895650791, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.6000218895650791, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 504.236, - "cuda_time_us": 212.47899999999998, - "pct_cuda_time": 2.9252030807841964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.166, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 366.506, - "cuda_time_us": 73.087, - "pct_cuda_time": 1.0061903414703317, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.856, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.2876756930146818, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.2876756930146818, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 96.48, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.0546275435433699, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.0546275435433699, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.563, - "cuda_time_us": 29.791, - "pct_cuda_time": 0.41013335425920683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.519, - "pct_cuda_time": 0.3237865918842028, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.034803031773598565, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 26.931, - "cuda_time_us": 18.432, - "pct_cuda_time": 0.2537537506530731, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.432, - "pct_cuda_time": 0.2537537506530731, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.794, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04537610471747661, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04537610471747661, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.713, - "cuda_time_us": 132.64, - "pct_cuda_time": 1.8260578063489372, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.361, - "cuda_time_us": 79.744, - "pct_cuda_time": 1.097837407339337, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.744, - "pct_cuda_time": 1.097837407339337, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.409, - "cuda_time_us": 9.184, - "pct_cuda_time": 0.1264363306205416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.184, - "pct_cuda_time": 0.1264363306205416, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.215, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6017840683890588, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6017840683890588, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 562.581, - "cuda_time_us": 212.222, - "pct_cuda_time": 2.921664956114175, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.867, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.049341007071430874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.049341007071430874, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 410.087, - "cuda_time_us": 72.224, - "pct_cuda_time": 0.9943094014305311, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.725, - "cuda_time_us": 20.32, - "pct_cuda_time": 0.27974588830677327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.32, - "pct_cuda_time": 0.27974588830677327, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.087, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.050662641189415644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.050662641189415644, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 221.741, - "cuda_time_us": 30.048, - "pct_cuda_time": 0.4136714789292285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171921883163414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.776, - "pct_cuda_time": 0.32732471655422446, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436248706760365, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.020265056475766253, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 32.57, - "cuda_time_us": 18.176, - "pct_cuda_time": 0.2502293930051137, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.176, - "pct_cuda_time": 0.2502293930051137, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.777, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.0462434271074041, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.0462434271074041, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.151, - "cuda_time_us": 133.055, - "pct_cuda_time": 1.831771120504809, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 38.339, - "cuda_time_us": 80.255, - "pct_cuda_time": 1.1048723556131934, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.255, - "pct_cuda_time": 1.1048723556131934, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.971, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12159033885459752, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12159033885459752, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.502, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6053084260370182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6053084260370182, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 496.491, - "cuda_time_us": 211.647, - "pct_cuda_time": 2.913748918428329, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.653, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 352.754, - "cuda_time_us": 72.896, - "pct_cuda_time": 1.0035608402564244, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.996, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2863540588966971, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2863540588966971, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.387, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 166.705, - "cuda_time_us": 30.112, - "pct_cuda_time": 0.41455256834121834, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.368, - "pct_cuda_time": 0.03260030824362397, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.552, - "pct_cuda_time": 0.32424090361226005, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.688, - "pct_cuda_time": 0.03700575530357316, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 26.815, - "cuda_time_us": 18.208, - "pct_cuda_time": 0.2506699377111087, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.208, - "pct_cuda_time": 0.2506699377111087, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.722, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.186, - "cuda_time_us": 131.839, - "pct_cuda_time": 1.8150304216770021, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.137, - "cuda_time_us": 79.423, - "pct_cuda_time": 1.0934181932573255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.423, - "pct_cuda_time": 1.0934181932573255, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.796, - "cuda_time_us": 8.833, - "pct_cuda_time": 0.12160410587665987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.833, - "pct_cuda_time": 0.12160410587665987, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.97, - "cuda_time_us": 43.583, - "pct_cuda_time": 0.6000081225430167, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.583, - "pct_cuda_time": 0.6000081225430167, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 507.974, - "cuda_time_us": 210.56199999999998, - "pct_cuda_time": 2.8988116994906887, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.439, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 358.786, - "cuda_time_us": 72.226, - "pct_cuda_time": 0.9943369354746558, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.451, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.135, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05154373060140547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05154373060140547, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 170.858, - "cuda_time_us": 29.889999999999997, - "pct_cuda_time": 0.41149628944337857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171921883163414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.585, - "pct_cuda_time": 0.32469521534031737, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436248706760365, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.020719368203823512, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 29.27, - "cuda_time_us": 17.952, - "pct_cuda_time": 0.24714558006314935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 17.952, - "pct_cuda_time": 0.24714558006314935, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.893, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.54, - "cuda_time_us": 131.456, - "pct_cuda_time": 1.809757652227125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.58, - "cuda_time_us": 79.264, - "pct_cuda_time": 1.0912292367494132, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.264, - "pct_cuda_time": 1.0912292367494132, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.923, - "cuda_time_us": 9.216, - "pct_cuda_time": 0.12687687532653655, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.216, - "pct_cuda_time": 0.12687687532653655, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.568, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.5916515401511756, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.5916515401511756, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 499.84, - "cuda_time_us": 211.933, - "pct_cuda_time": 2.9176862867381583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.661, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 359.251, - "cuda_time_us": 73.086, - "pct_cuda_time": 1.0061765744482694, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.694, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2863540588966971, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2863540588966971, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 110.097, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.051103185895410554, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.051103185895410554, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.835, - "cuda_time_us": 30.078999999999997, - "pct_cuda_time": 0.4140982566131611, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.615, - "pct_cuda_time": 0.32510822600218753, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.72, - "pct_cuda_time": 0.03744630000956808, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.002, - "cuda_time_us": 18.495, - "pct_cuda_time": 0.25462107304300063, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.495, - "pct_cuda_time": 0.25462107304300063, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.287, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 92.025, - "cuda_time_us": 131.80700000000002, - "pct_cuda_time": 1.8145898769710074, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.008, - "cuda_time_us": 79.775, - "pct_cuda_time": 1.0982641850232697, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.775, - "pct_cuda_time": 1.0982641850232697, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.151, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.11806598120663817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.11806598120663817, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.545, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.5982597107410994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.5982597107410994, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 472.827, - "cuda_time_us": 211.77700000000002, - "pct_cuda_time": 2.9155386312964335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.771, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 336.553, - "cuda_time_us": 72.929, - "pct_cuda_time": 1.0040151519844818, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.469, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.28547296948470724, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.28547296948470724, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 96.158, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.05330590942538515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.05330590942538515, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.534, - "cuda_time_us": 29.857, - "pct_cuda_time": 0.41104197771532136, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.305, - "pct_cuda_time": 0.03173298585369648, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.32468144831825496, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436248706760365, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.020265056475766253, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 26.554, - "cuda_time_us": 18.464, - "pct_cuda_time": 0.25419429535906796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.464, - "pct_cuda_time": 0.25419429535906796, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.557, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 90.748, - "cuda_time_us": 131.96800000000002, - "pct_cuda_time": 1.8168063675230444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 31.906, - "cuda_time_us": 79.808, - "pct_cuda_time": 1.098718496751327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.808, - "pct_cuda_time": 1.098718496751327, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.123, - "cuda_time_us": 8.607, - "pct_cuda_time": 0.11849275889057075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.607, - "pct_cuda_time": 0.11849275889057075, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 27.978, - "cuda_time_us": 43.553, - "pct_cuda_time": 0.5995951118811464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.553, - "pct_cuda_time": 0.5995951118811464, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 465.851, - "cuda_time_us": 212.09299999999996, - "pct_cuda_time": 2.9198890102681325, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.815, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.049767784755363455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.049767784755363455, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 327.601, - "cuda_time_us": 72.606, - "pct_cuda_time": 0.9995684038583456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.067, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.281508067130753, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.281508067130753, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.273, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 150.169, - "cuda_time_us": 30.207, - "pct_cuda_time": 0.4158604354371408, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03215976353762905, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.68, - "pct_cuda_time": 0.32600308243623977, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.687, - "pct_cuda_time": 0.03699198828151082, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 25.891, - "cuda_time_us": 18.175, - "pct_cuda_time": 0.2502156259830514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.175, - "pct_cuda_time": 0.2502156259830514, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.621, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.178, - "cuda_time_us": 132.54399999999998, - "pct_cuda_time": 1.8247361722309525, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.669, - "cuda_time_us": 80.0, - "pct_cuda_time": 1.1013617649872964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.0, - "pct_cuda_time": 1.1013617649872964, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.37, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.11806598120663817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.11806598120663817, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.872, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6053084260370182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6053084260370182, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 499.384, - "cuda_time_us": 211.425, - "pct_cuda_time": 2.9106926395304895, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.364, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.047138283541456286, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 359.791, - "cuda_time_us": 71.937, - "pct_cuda_time": 0.9903582660986392, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.858, - "cuda_time_us": 20.383, - "pct_cuda_time": 0.28061321069670075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.383, - "pct_cuda_time": 0.28061321069670075, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.205, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 172.066, - "cuda_time_us": 29.698, - "pct_cuda_time": 0.4088530212074091, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03215976353762905, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.425, - "pct_cuda_time": 0.32249249181034273, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.465, - "pct_cuda_time": 0.03393570938367107, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.020265056475766253, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.625, - "cuda_time_us": 18.24, - "pct_cuda_time": 0.25111048241710354, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.24, - "pct_cuda_time": 0.25111048241710354, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.23, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.501, - "cuda_time_us": 132.73600000000002, - "pct_cuda_time": 1.8273794404669226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.611, - "cuda_time_us": 80.128, - "pct_cuda_time": 1.103123943811276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.128, - "pct_cuda_time": 1.103123943811276, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.036, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12159033885459752, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12159033885459752, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.716, - "cuda_time_us": 43.776, - "pct_cuda_time": 0.6026651578010487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.776, - "pct_cuda_time": 0.6026651578010487, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 562.585, - "cuda_time_us": 211.138, - "pct_cuda_time": 2.9067415041985973, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.004, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04669773883546136, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04669773883546136, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 422.281, - "cuda_time_us": 73.05799999999999, - "pct_cuda_time": 1.0057910978305238, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.108, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841513353667225, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.573, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.05418699883737499, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.05418699883737499, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 233.512, - "cuda_time_us": 29.889999999999997, - "pct_cuda_time": 0.41149628944337857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.648, - "pct_cuda_time": 0.32556253773024485, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.497, - "pct_cuda_time": 0.034376254089665985, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.020719368203823512, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 28.396, - "cuda_time_us": 18.592, - "pct_cuda_time": 0.2559564741830477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.592, - "pct_cuda_time": 0.2559564741830477, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.499, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.996, - "cuda_time_us": 131.36, - "pct_cuda_time": 1.808436018109141, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.486, - "cuda_time_us": 79.648, - "pct_cuda_time": 1.0965157732213522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.648, - "pct_cuda_time": 1.0965157732213522, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.3, - "cuda_time_us": 8.512, - "pct_cuda_time": 0.11718489179464835, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.512, - "pct_cuda_time": 0.11718489179464835, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.743, - "cuda_time_us": 43.2, - "pct_cuda_time": 0.5947353530931401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.2, - "pct_cuda_time": 0.5947353530931401, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 494.003, - "cuda_time_us": 212.735, - "pct_cuda_time": 2.928727438432156, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.805, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 351.103, - "cuda_time_us": 74.239, - "pct_cuda_time": 1.0220499508861487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 26.62, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.29472440831060054, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.29472440831060054, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.807, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.049767784755363455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.049767784755363455, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 168.468, - "cuda_time_us": 29.983999999999998, - "pct_cuda_time": 0.4127903895172386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171921883163414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.456, - "pct_cuda_time": 0.3229192694942753, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.72, - "pct_cuda_time": 0.03744630000956808, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 28.851, - "cuda_time_us": 19.232, - "pct_cuda_time": 0.26476736830294606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 19.232, - "pct_cuda_time": 0.26476736830294606, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.192, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.399, - "cuda_time_us": 131.55200000000002, - "pct_cuda_time": 1.8110792863451106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.901, - "cuda_time_us": 79.712, - "pct_cuda_time": 1.0973968626333421, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.712, - "pct_cuda_time": 1.0973968626333421, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.941, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.163, - "cuda_time_us": 43.2, - "pct_cuda_time": 0.5947353530931401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.2, - "pct_cuda_time": 0.5947353530931401, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 511.068, - "cuda_time_us": 213.28099999999998, - "pct_cuda_time": 2.9362442324781943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.925, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 369.334, - "cuda_time_us": 73.69699999999999, - "pct_cuda_time": 1.0145882249283595, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.647, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.29736767654657004, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.29736767654657004, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 118.266, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 165.676, - "cuda_time_us": 29.824999999999996, - "pct_cuda_time": 0.41060143300932644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.52, - "pct_cuda_time": 0.32380035890626513, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.035684121185588405, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.473, - "pct_cuda_time": 0.020278823497828596, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.38, - "cuda_time_us": 18.496, - "pct_cuda_time": 0.2546348400650629, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.496, - "pct_cuda_time": 0.2546348400650629, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.391, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.509, - "cuda_time_us": 132.48, - "pct_cuda_time": 1.8238550828189628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.995, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1053266673412505, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1053266673412505, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.668, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.672, - "cuda_time_us": 43.552, - "pct_cuda_time": 0.5995813448590842, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.552, - "pct_cuda_time": 0.5995813448590842, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 541.273, - "cuda_time_us": 211.26, - "pct_cuda_time": 2.9084210808902027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.488, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 385.782, - "cuda_time_us": 72.733, - "pct_cuda_time": 1.001316815660263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.487, - "cuda_time_us": 20.352, - "pct_cuda_time": 0.2801864330127682, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.352, - "pct_cuda_time": 0.2801864330127682, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 112.447, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 177.292, - "cuda_time_us": 30.43, - "pct_cuda_time": 0.41893048135704286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.656, - "pct_cuda_time": 0.03656521059757824, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.775, - "pct_cuda_time": 0.3273109495321621, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.495, - "pct_cuda_time": 0.034348720045541306, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 37.39, - "cuda_time_us": 18.335, - "pct_cuda_time": 0.252418349513026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.335, - "pct_cuda_time": 0.252418349513026, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.206, - "cuda_time_us": 3.423, - "pct_cuda_time": 0.04712451651939394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.423, - "pct_cuda_time": 0.04712451651939394, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 105.583, - "cuda_time_us": 131.648, - "pct_cuda_time": 1.8124009204630949, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.715, - "cuda_time_us": 79.2, - "pct_cuda_time": 1.0903481473374235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.2, - "pct_cuda_time": 1.0903481473374235, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.512, - "cuda_time_us": 9.6, - "pct_cuda_time": 0.13216341179847554, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.6, - "pct_cuda_time": 0.13216341179847554, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.204, - "cuda_time_us": 42.848, - "pct_cuda_time": 0.5898893613271959, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.848, - "pct_cuda_time": 0.5898893613271959, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 515.543, - "cuda_time_us": 211.70800000000003, - "pct_cuda_time": 2.9145887067741323, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.087, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 370.764, - "cuda_time_us": 72.893, - "pct_cuda_time": 1.0035195391902376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.42, - "cuda_time_us": 20.703, - "pct_cuda_time": 0.28501865775665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.703, - "pct_cuda_time": 0.28501865775665, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.206, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.0546275435433699, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.0546275435433699, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 169.081, - "cuda_time_us": 29.726999999999997, - "pct_cuda_time": 0.40925226484721694, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.423, - "pct_cuda_time": 0.32246495776621803, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03524357647959349, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 29.092, - "cuda_time_us": 18.495, - "pct_cuda_time": 0.25462107304300063, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.495, - "pct_cuda_time": 0.25462107304300063, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.174, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.46, - "cuda_time_us": 132.127, - "pct_cuda_time": 1.8189953240309567, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.892, - "cuda_time_us": 79.487, - "pct_cuda_time": 1.0942992826693154, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.487, - "pct_cuda_time": 1.0942992826693154, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.915, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114979414860261, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114979414860261, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.121, - "cuda_time_us": 43.84, - "pct_cuda_time": 0.6035462472130384, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.84, - "pct_cuda_time": 0.6035462472130384, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 480.118, - "cuda_time_us": 211.74400000000003, - "pct_cuda_time": 2.9150843195683764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.279, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 342.387, - "cuda_time_us": 72.063, - "pct_cuda_time": 0.9920929108784943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.534, - "cuda_time_us": 20.416, - "pct_cuda_time": 0.281067522424758, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.416, - "pct_cuda_time": 0.281067522424758, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.099, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.052865364719390226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.052865364719390226, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.056, - "cuda_time_us": 29.759, - "pct_cuda_time": 0.4096928095532119, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921942361608726, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.263, - "pct_cuda_time": 0.3202622342362435, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03524357647959349, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.020265056475766253, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.424, - "cuda_time_us": 18.048, - "pct_cuda_time": 0.24846721418113402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.048, - "pct_cuda_time": 0.24846721418113402, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.258, - "cuda_time_us": 3.297, - "pct_cuda_time": 0.045389871739538956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.297, - "pct_cuda_time": 0.045389871739538956, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.337, - "cuda_time_us": 132.89600000000002, - "pct_cuda_time": 1.829582163996897, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.284, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1053266673412505, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1053266673412505, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.636, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.1185065259126331, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.1185065259126331, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.066, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.605748970743013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.605748970743013, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 481.054, - "cuda_time_us": 210.942, - "pct_cuda_time": 2.9040431678743786, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.208, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.049341007071430874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.049341007071430874, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 337.538, - "cuda_time_us": 72.672, - "pct_cuda_time": 1.00047702731446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.816, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.2876756930146818, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.2876756930146818, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 96.11, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05154373060140547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05154373060140547, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.108, - "cuda_time_us": 29.696, - "pct_cuda_time": 0.40882548716328443, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.488, - "pct_cuda_time": 0.3233598142002702, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921942361608726, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.663, - "cuda_time_us": 18.336, - "pct_cuda_time": 0.2524321165350883, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.336, - "pct_cuda_time": 0.2524321165350883, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.857, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04493556001148169, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04493556001148169, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.212, - "cuda_time_us": 131.422, - "pct_cuda_time": 1.8092895734770058, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.644, - "cuda_time_us": 79.775, - "pct_cuda_time": 1.0982641850232697, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.775, - "pct_cuda_time": 1.0982641850232697, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.397, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12247142826658738, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12247142826658738, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.535, - "cuda_time_us": 42.751, - "pct_cuda_time": 0.5885539601871488, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.751, - "pct_cuda_time": 0.5885539601871488, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 565.151, - "cuda_time_us": 210.784, - "pct_cuda_time": 2.9018679783885286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.256, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 414.819, - "cuda_time_us": 72.16, - "pct_cuda_time": 0.9934283120185412, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.788, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.281508067130753, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.281508067130753, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 110.451, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.050662641189415644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.050662641189415644, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 215.644, - "cuda_time_us": 29.759999999999998, - "pct_cuda_time": 0.4097065765752742, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436248706760365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.136, - "pct_cuda_time": 0.31851382243432613, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921942361608726, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.022908324711735765, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 29.573, - "cuda_time_us": 18.272, - "pct_cuda_time": 0.25155102712309846, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.272, - "pct_cuda_time": 0.25155102712309846, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.856, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.046257194129466446, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 105.03, - "cuda_time_us": 131.648, - "pct_cuda_time": 1.8124009204630949, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 40.125, - "cuda_time_us": 79.84, - "pct_cuda_time": 1.0991590414573218, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.84, - "pct_cuda_time": 1.0991590414573218, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.768, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114979414860261, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114979414860261, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.112, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.5920920848571706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.5920920848571706, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 484.206, - "cuda_time_us": 211.551, - "pct_cuda_time": 2.9124272843103443, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.696, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 343.371, - "cuda_time_us": 72.83099999999999, - "pct_cuda_time": 1.0026659838223722, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.056, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.2850324247787123, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.2850324247787123, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.733, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.962, - "cuda_time_us": 29.822999999999997, - "pct_cuda_time": 0.4105738989652017, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.031278674125639214, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.519, - "pct_cuda_time": 0.3237865918842028, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.034803031773598565, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 25.962, - "cuda_time_us": 18.528, - "pct_cuda_time": 0.25507538477105784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.528, - "pct_cuda_time": 0.25507538477105784, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.608, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.576, - "cuda_time_us": 131.904, - "pct_cuda_time": 1.8159252781110542, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.763, - "cuda_time_us": 79.776, - "pct_cuda_time": 1.0982779520453319, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.776, - "pct_cuda_time": 1.0982779520453319, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 19.99, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511469650255685, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511469650255685, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.77, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.5925326295631654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.5925326295631654, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 483.664, - "cuda_time_us": 212.127, - "pct_cuda_time": 2.920357089018253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.093, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845991765944104, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 342.872, - "cuda_time_us": 72.512, - "pct_cuda_time": 0.9982743037844855, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.648, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.2806269777187631, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.2806269777187631, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.109, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.049341007071430874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.049341007071430874, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.672, - "cuda_time_us": 30.144, - "pct_cuda_time": 0.4149931130472132, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171921883163414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.776, - "pct_cuda_time": 0.32732471655422446, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03524357647959349, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 26.283, - "cuda_time_us": 18.4, - "pct_cuda_time": 0.2533132059470782, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.4, - "pct_cuda_time": 0.2533132059470782, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.167, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.522, - "cuda_time_us": 132.607, - "pct_cuda_time": 1.8256034946208803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.51, - "cuda_time_us": 80.639, - "pct_cuda_time": 1.1101588920851324, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.639, - "pct_cuda_time": 1.1101588920851324, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.741, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894707061862803, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.238, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.5964975319171197, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.5964975319171197, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 520.5, - "cuda_time_us": 211.358, - "pct_cuda_time": 2.9097702490523125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.423, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.048900462365435965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.048900462365435965, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 378.082, - "cuda_time_us": 72.92699999999999, - "pct_cuda_time": 1.003987617940357, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.63, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2837107906607276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2837107906607276, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.265, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 170.257, - "cuda_time_us": 29.823, - "pct_cuda_time": 0.4105738989652018, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.033040852949618886, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.393, - "pct_cuda_time": 0.3220519471043478, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.527, - "pct_cuda_time": 0.03478926475153623, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.02069183415969883, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 49.754, - "cuda_time_us": 18.72, - "pct_cuda_time": 0.25771865300702734, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.72, - "pct_cuda_time": 0.25771865300702734, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.245, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581664942347153, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.28, - "cuda_time_us": 131.551, - "pct_cuda_time": 1.8110655193230476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.124, - "cuda_time_us": 79.968, - "pct_cuda_time": 1.1009212202813015, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.968, - "pct_cuda_time": 1.1009212202813015, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.274, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.11982816003061784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.11982816003061784, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.735, - "cuda_time_us": 42.879, - "pct_cuda_time": 0.5903161390111284, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.879, - "pct_cuda_time": 0.5903161390111284, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 514.753, - "cuda_time_us": 212.28699999999998, - "pct_cuda_time": 2.922559812548227, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.005, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 363.538, - "cuda_time_us": 73.27799999999999, - "pct_cuda_time": 1.0088198426842387, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.737, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.2929622294866209, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.2929622294866209, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.275, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.049781551777425805, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.711, - "cuda_time_us": 30.046999999999997, - "pct_cuda_time": 0.4136577119071661, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921942361608726, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.519, - "pct_cuda_time": 0.3237865918842028, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03524357647959349, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.020705601181761173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 28.658, - "cuda_time_us": 18.335, - "pct_cuda_time": 0.252418349513026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.335, - "pct_cuda_time": 0.252418349513026, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.325, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.046711505857523705, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.046711505857523705, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.35, - "cuda_time_us": 132.0, - "pct_cuda_time": 1.817246912229039, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.599, - "cuda_time_us": 79.904, - "pct_cuda_time": 1.1000401308693115, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.904, - "pct_cuda_time": 1.1000401308693115, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.917, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12291197297258229, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12291197297258229, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 35.428, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.5942948083871451, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.5942948083871451, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 525.645, - "cuda_time_us": 211.199, - "pct_cuda_time": 2.9075812925444002, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.532, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019372953446125, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 376.326, - "cuda_time_us": 72.607, - "pct_cuda_time": 0.9995821708804078, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.327, - "cuda_time_us": 20.703, - "pct_cuda_time": 0.28501865775665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.703, - "pct_cuda_time": 0.28501865775665, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.715, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.051984275307400386, - "trace": "_C::rotary_embedding(int64[1], bfloat16[1, 4096], bfloat16[1, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 173.146, - "cuda_time_us": 30.016, - "pct_cuda_time": 0.4132309342232336, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03215976353762905, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.648, - "pct_cuda_time": 0.32556253773024485, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.527, - "pct_cuda_time": 0.03478926475153623, - "trace": "_vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.020719368203823512, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[1, 1, 32, 128], None, None, None, None, int32[1], None, None, int32[1, 257], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1, 32, 128], bfloat16[1, 8, 128], bfloat16[1, 8, 128], bfloat16[1, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 27.844, - "cuda_time_us": 18.112, - "pct_cuda_time": 0.24934830359312388, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void gemv2T_kernel_val, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float> >(cublasGemvParamsEx, cublasGemvTensorStridedBatched<__nv_bfloat16 const>, cublasGemvTensorStridedBatched<__nv_bfloat16>, float>, float, float)", - "cpu_time_us": 0, - "cuda_time_us": 18.112, - "pct_cuda_time": 0.24934830359312388, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.427, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.048900462365435965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.048900462365435965, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 103.171, - "cuda_time_us": 131.55200000000002, - "pct_cuda_time": 1.8110792863451106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.067, - "cuda_time_us": 78.816, - "pct_cuda_time": 1.0850616108654845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 78.816, - "pct_cuda_time": 1.0850616108654845, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.306, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12026870473661277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12026870473661277, - "trace": "_C::silu_and_mul(bfloat16[1, 14336], bfloat16[1, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 36.455, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.605748970743013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.605748970743013, - "trace": "mm(bfloat16[1, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.235, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0475788282474512, - "trace": "_C::fused_add_rms_norm(bfloat16[1, 4096], bfloat16[1, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 94.129, - "cuda_time_us": 347.518, - "pct_cuda_time": 4.784287973060691, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 2.624, - "pct_cuda_time": 0.03612466589158332, - "trace": "index_select(bfloat16[1, 4096], 0, int64[1])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.010118761215820785, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[1, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 344.159, - "pct_cuda_time": 4.7380445459532865, - "trace": "mm(bfloat16[1, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[1, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[1, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 684.941, - "cuda_time_us": 124.28700000000002, - "pct_cuda_time": 1.7110618710622016, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.815, - "pct_cuda_time": 0.03875416710549049, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03039758471364938, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03039758471364938, - "trace": "copy_(int32[1], int32[1], True) <- _to_copy(int32[1], 3, 0, None, None, True, None) <- to(int32[1], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03039758471364938, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381294196443, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03039758471364938, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.02951649530165954, - "trace": "copy_(bfloat16[1], bfloat16[1], True) <- _to_copy(bfloat16[1], 15, 0, None, None, True, None) <- to(bfloat16[1], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.192, - "pct_cuda_time": 0.057711356485334334, - "trace": "copy_(float32[1, 128256], bfloat16[1, 128256], False) <- _to_copy(bfloat16[1, 128256], 6, None, None, None, False, None) <- to(bfloat16[1, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.062116803545283504, - "trace": "div_(float32[1, 128256], bfloat16[1, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.08, - "pct_cuda_time": 0.46918011188458825, - "trace": "_softmax(float32[1, 128256], -1, False) <- softmax(float32[1, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.584, - "pct_cuda_time": 0.3797495365676198, - "trace": "_log_softmax(float32[1, 128256], -1, False) <- log_softmax(float32[1, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 1.888, - "pct_cuda_time": 0.025992137653700193, - "trace": "copy_(int64[1], int32[1], False) <- _to_copy(int32[1], 4, None, None, None, False, None) <- to(int32[1], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.06387898236926319, - "trace": "index(float32[1, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 28.576, - "pct_cuda_time": 0.3934064224534623, - "trace": "argmax(float32[1, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.784, - "pct_cuda_time": 0.03832738942155792, - "trace": "copy_(int64[1], int64[1], False) <- _to_copy(int64[1], 4, 0, None, None, False, None) <- to(int64[1], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - } -} \ No newline at end of file diff --git a/H100_llama8b_pp1_tp1/profiling_bs2_pl1024.json b/H100_llama8b_pp1_tp1/profiling_bs2_pl1024.json deleted file mode 100644 index c2aeeda2804d580410bf958172594669ac911db4..0000000000000000000000000000000000000000 --- a/H100_llama8b_pp1_tp1/profiling_bs2_pl1024.json +++ /dev/null @@ -1,18548 +0,0 @@ -{ - "context": { - "python_version": "3.12.9 | packaged by Anaconda, Inc. | (main, Feb 6 2025, 18:56:27) [GCC 11.2.0]", - "torch_version": "2.5.1+cu124", - "engine_args": { - "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "served_model_name": null, - "tokenizer": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "task": "auto", - "skip_tokenizer_init": false, - "tokenizer_mode": "auto", - "trust_remote_code": false, - "allowed_local_media_path": null, - "download_dir": null, - "load_format": "dummy", - "config_format": "auto", - "dtype": "auto", - "kv_cache_dtype": "auto", - "seed": 0, - "max_model_len": null, - "distributed_executor_backend": null, - "pipeline_parallel_size": 1, - "tensor_parallel_size": 1, - "max_parallel_loading_workers": null, - "block_size": null, - "enable_prefix_caching": false, - "disable_sliding_window": false, - "use_v2_block_manager": true, - "swap_space": 4, - "cpu_offload_gb": 0, - "gpu_memory_utilization": 0.9, - "max_num_batched_tokens": 8000, - "max_num_partial_prefills": 1, - "max_long_partial_prefills": 1, - "long_prefill_token_threshold": 0, - "max_num_seqs": 256, - "max_logprobs": 20, - "disable_log_stats": false, - "revision": null, - "code_revision": null, - "rope_scaling": null, - "rope_theta": null, - "hf_overrides": null, - "tokenizer_revision": null, - "quantization": null, - "enforce_eager": true, - "max_seq_len_to_capture": 8192, - "disable_custom_all_reduce": false, - "tokenizer_pool_size": 0, - "tokenizer_pool_type": "ray", - "tokenizer_pool_extra_config": null, - "limit_mm_per_prompt": null, - "mm_processor_kwargs": null, - "disable_mm_preprocessor_cache": false, - "enable_lora": false, - "enable_lora_bias": false, - "max_loras": 1, - "max_lora_rank": 16, - "enable_prompt_adapter": false, - "max_prompt_adapters": 1, - "max_prompt_adapter_token": 0, - "fully_sharded_loras": false, - "lora_extra_vocab_size": 256, - "long_lora_scaling_factors": null, - "lora_dtype": "auto", - "max_cpu_loras": null, - "device": "auto", - "num_scheduler_steps": 1, - "multi_step_stream_outputs": true, - "ray_workers_use_nsight": false, - "num_gpu_blocks_override": null, - "num_lookahead_slots": 0, - "model_loader_extra_config": null, - "ignore_patterns": [], - "preemption_mode": null, - "scheduler_delay_factor": 0.0, - "enable_chunked_prefill": null, - "guided_decoding_backend": "xgrammar", - "logits_processor_pattern": null, - "speculative_model": null, - "speculative_model_quantization": null, - "speculative_draft_tensor_parallel_size": null, - "num_speculative_tokens": null, - "speculative_disable_mqa_scorer": false, - "speculative_max_model_len": null, - "speculative_disable_by_batch_size": null, - "ngram_prompt_lookup_max": null, - "ngram_prompt_lookup_min": null, - "spec_decoding_acceptance_method": "rejection_sampler", - "typical_acceptance_sampler_posterior_threshold": null, - "typical_acceptance_sampler_posterior_alpha": null, - "qlora_adapter_name_or_path": null, - "disable_logprobs_during_spec_decoding": null, - "otlp_traces_endpoint": null, - "collect_detailed_traces": null, - "disable_async_output_proc": false, - "scheduling_policy": "fcfs", - "scheduler_cls": "vllm.core.scheduler.Scheduler", - "override_neuron_config": null, - "override_pooler_config": null, - "compilation_config": null, - "worker_cls": "auto", - "kv_transfer_config": null, - "generation_config": null, - "override_generation_config": null, - "enable_sleep_mode": false, - "model_impl": "auto", - "calculate_kv_scales": false, - "additional_config": null - }, - "prompt_len": 0, - "batch_size": 2, - "num_steps": 2, - "complete_num_requests_per_step": null, - "save_chrome_traces_folder": null - }, - "prefill": { - "metadata": { - "num_running_seqs": null - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 44809.865999999995, - "pct_cuda_time": 98.91136689472341, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 58.624, - "pct_cuda_time": 0.12940409089454244, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cuda_time_us": 58.624, - "pct_cuda_time": 0.12940409089454244, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 44730.18599999999, - "pct_cuda_time": 98.73548469694643, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 1338.849, - "pct_cuda_time": 2.9553175779555683, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 27.136, - "pct_cuda_time": 0.05989883683328165, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 1311.713, - "pct_cuda_time": 2.895418741122287, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 11173.193, - "pct_cuda_time": 24.663224661474228, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 4704.341, - "pct_cuda_time": 10.384159565415576, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 32.28500000000001, - "pct_cuda_time": 0.07126451751041048, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 4672.056, - "pct_cuda_time": 10.312895047905164, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 828.1880000000001, - "pct_cuda_time": 1.8281064961409885, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 828.1880000000001, - "pct_cuda_time": 1.8281064961409885, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 2142.364, - "pct_cuda_time": 4.728961957307511, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 368.607, - "pct_cuda_time": 0.813647204768774, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 1727.4859999999999, - "pct_cuda_time": 3.8131781414275636, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 46.271, - "pct_cuda_time": 0.10213661111117243, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 3498.2999999999997, - "pct_cuda_time": 7.721996642610156, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 32.738, - "pct_cuda_time": 0.07226445018602501, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 3465.562000000001, - "pct_cuda_time": 7.649732192424133, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 32218.144000000004, - "pct_cuda_time": 71.11694245751667, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 19929.113, - "pct_cuda_time": 43.990665087670706, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 31.90900000000001, - "pct_cuda_time": 0.0704345513160814, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 19897.203999999998, - "pct_cuda_time": 43.920230536354616, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 2862.1070000000004, - "pct_cuda_time": 6.3176916344484555, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 2862.1070000000004, - "pct_cuda_time": 6.3176916344484555, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 9426.923999999999, - "pct_cuda_time": 20.808585735397507, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 9426.923999999999, - "pct_cuda_time": 20.808585735397507, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 21.056, - "pct_cuda_time": 0.046478106882428455, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 21.056, - "pct_cuda_time": 0.046478106882428455, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 357.375, - "pct_cuda_time": 0.7888541720700925, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.008, - "pct_cuda_time": 0.006639729554632635, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 353.311, - "pct_cuda_time": 0.7798834736292589, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 135.809, - "pct_cuda_time": 0.2997789332064839, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 16.48, - "pct_cuda_time": 0.03637724170889157, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.448, - "pct_cuda_time": 0.009818323490361026, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 5.088, - "pct_cuda_time": 0.01123103190624031, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 37.664, - "pct_cuda_time": 0.08313789027449588, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 31.008, - "pct_cuda_time": 0.06844572274935132, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 2.176, - "pct_cuda_time": 0.004803208613989567, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 5.184, - "pct_cuda_time": 0.011442938168622203, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 30.656, - "pct_cuda_time": 0.0676687331206177, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 3.105, - "pct_cuda_time": 0.00685384317391434, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 17938.734, - "cuda_time_us": 44809.865999999995, - "pct_cuda_time": 98.91136689472341, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 63.065, - "cuda_time_us": 58.624, - "pct_cuda_time": 0.12940409089454244, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 58.624, - "pct_cuda_time": 0.12940409089454244, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[2048]) <- embedding(bfloat16[128256, 4096], int64[2048], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 874.014, - "cuda_time_us": 1409.21, - "pct_cuda_time": 3.110629416783197, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 58.053, - "cuda_time_us": 27.136, - "pct_cuda_time": 0.05989883683328165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.136, - "pct_cuda_time": 0.05989883683328165, - "trace": "_C::rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 628.647, - "cuda_time_us": 357.182, - "pct_cuda_time": 0.7884281521884289, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 85.234, - "cuda_time_us": 152.703, - "pct_cuda_time": 0.3370700206718974, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0016224073213613654, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 151.968, - "pct_cuda_time": 0.335447613350536, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 191.572, - "cuda_time_us": 25.632, - "pct_cuda_time": 0.056578972055965335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.632, - "pct_cuda_time": 0.056578972055965335, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 227.05, - "cuda_time_us": 67.19999999999999, - "pct_cuda_time": 0.14833438366732482, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.025781928589796936, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.176, - "pct_cuda_time": 0.11958576740418142, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.39, - "cuda_time_us": 111.647, - "pct_cuda_time": 0.24644477579324134, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 110.911, - "pct_cuda_time": 0.24482016111498014, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.966, - "cuda_time_us": 21.087, - "pct_cuda_time": 0.046546534946322604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.087, - "pct_cuda_time": 0.046546534946322604, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 128.122, - "cuda_time_us": 1003.8050000000001, - "pct_cuda_time": 2.215755892815164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 46.279, - "cuda_time_us": 620.541, - "pct_cuda_time": 1.3697554579658546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0016224073213613654, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 619.806, - "pct_cuda_time": 1.3681330506444933, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 29.116, - "cuda_time_us": 89.696, - "pct_cuda_time": 0.1979910844854817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.696, - "pct_cuda_time": 0.1979910844854817, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 36.4, - "cuda_time_us": 293.568, - "pct_cuda_time": 0.6480093503638276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.568, - "pct_cuda_time": 0.6480093503638276, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 582.233, - "cuda_time_us": 1395.999, - "pct_cuda_time": 3.081468024779789, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.375, - "cuda_time_us": 19.777, - "pct_cuda_time": 0.04365489740756969, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.777, - "pct_cuda_time": 0.04365489740756969, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 426.032, - "cuda_time_us": 349.568, - "pct_cuda_time": 0.7716213367532649, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 35.801, - "cuda_time_us": 146.304, - "pct_cuda_time": 0.3229451438700044, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0022603334654068548, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.28, - "pct_cuda_time": 0.3206848104045975, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 129.301, - "cuda_time_us": 26.304, - "pct_cuda_time": 0.05806231589263858, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.304, - "pct_cuda_time": 0.05806231589263858, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 174.74, - "cuda_time_us": 67.52000000000001, - "pct_cuda_time": 0.1490407378752645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.025570022327415044, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.176, - "pct_cuda_time": 0.11958576740418142, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.76, - "pct_cuda_time": 0.0038849481436680315, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.239, - "cuda_time_us": 109.44, - "pct_cuda_time": 0.2415731391153576, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.096, - "pct_cuda_time": 0.2386064514420111, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.001, - "cuda_time_us": 21.44, - "pct_cuda_time": 0.04732573193195602, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.44, - "pct_cuda_time": 0.04732573193195602, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.321, - "cuda_time_us": 1005.214, - "pct_cuda_time": 2.218866058686998, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.437, - "cuda_time_us": 621.214, - "pct_cuda_time": 1.3712410091594276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0016224073213613654, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 620.479, - "pct_cuda_time": 1.3696186018380663, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.53, - "cuda_time_us": 89.472, - "pct_cuda_time": 0.19749663653992394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.472, - "pct_cuda_time": 0.19749663653992394, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.168, - "cuda_time_us": 294.528, - "pct_cuda_time": 0.6501284129876467, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.528, - "pct_cuda_time": 0.6501284129876467, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 530.847, - "cuda_time_us": 1396.51, - "pct_cuda_time": 3.0825959841555926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.243, - "cuda_time_us": 19.84, - "pct_cuda_time": 0.04379396089225781, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.84, - "pct_cuda_time": 0.04379396089225781, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 377.476, - "cuda_time_us": 348.607, - "pct_cuda_time": 0.7695000667725463, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.133, - "cuda_time_us": 146.975, - "pct_cuda_time": 0.3244262803497778, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.0021896980446128905, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.983, - "pct_cuda_time": 0.32223658230516494, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.786, - "cuda_time_us": 25.44, - "pct_cuda_time": 0.05615515953120155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.44, - "pct_cuda_time": 0.05615515953120155, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.216, - "cuda_time_us": 66.88000000000001, - "pct_cuda_time": 0.14762802945938522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.025640657748209006, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.76, - "pct_cuda_time": 0.11866750693385987, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0033198647773163177, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 45.993, - "cuda_time_us": 109.312, - "pct_cuda_time": 0.2412905974321817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0022603334654068548, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.288, - "pct_cuda_time": 0.23903026396677488, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.76, - "cuda_time_us": 21.505, - "pct_cuda_time": 0.04746921013044376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.505, - "pct_cuda_time": 0.04746921013044376, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.969, - "cuda_time_us": 1006.558, - "pct_cuda_time": 2.2218327463603447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.951, - "cuda_time_us": 623.1669999999999, - "pct_cuda_time": 1.375551977184759, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.001626822035160988, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.43, - "pct_cuda_time": 1.3739251551495981, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.352, - "cuda_time_us": 89.024, - "pct_cuda_time": 0.19650774064880844, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.024, - "pct_cuda_time": 0.19650774064880844, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.475, - "cuda_time_us": 294.367, - "pct_cuda_time": 0.649773028526777, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.367, - "pct_cuda_time": 0.649773028526777, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 520.759, - "cuda_time_us": 1397.981, - "pct_cuda_time": 3.085843006155215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.233, - "cuda_time_us": 19.999, - "pct_cuda_time": 0.04414493063932782, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.999, - "pct_cuda_time": 0.04414493063932782, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 370.326, - "cuda_time_us": 347.648, - "pct_cuda_time": 0.7673832115056272, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.421, - "cuda_time_us": 146.144, - "pct_cuda_time": 0.3225919667660346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0022603334654068548, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.12, - "pct_cuda_time": 0.3203316333006277, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.088, - "cuda_time_us": 26.304, - "pct_cuda_time": 0.05806231589263858, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.304, - "pct_cuda_time": 0.05806231589263858, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.322, - "cuda_time_us": 66.68799999999999, - "pct_cuda_time": 0.14720421693462138, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.392, - "pct_cuda_time": 0.025146209802651257, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.952, - "pct_cuda_time": 0.11909131945862365, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.51, - "cuda_time_us": 108.512, - "pct_cuda_time": 0.23952471191233263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.776, - "pct_cuda_time": 0.23790009723407146, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.966, - "cuda_time_us": 21.344, - "pct_cuda_time": 0.04711382566957413, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.344, - "pct_cuda_time": 0.04711382566957413, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.86, - "cuda_time_us": 1008.99, - "pct_cuda_time": 2.227201038340686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.013, - "cuda_time_us": 624.6389999999999, - "pct_cuda_time": 1.3788012065412814, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.377, - "pct_cuda_time": 0.0030395304510402728, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 623.262, - "pct_cuda_time": 1.3757616760902414, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.897, - "cuda_time_us": 89.248, - "pct_cuda_time": 0.1970021885943662, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.248, - "pct_cuda_time": 0.1970021885943662, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.851, - "cuda_time_us": 295.103, - "pct_cuda_time": 0.6513976432050382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.103, - "pct_cuda_time": 0.6513976432050382, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 497.621, - "cuda_time_us": 1397.949, - "pct_cuda_time": 3.085772370734421, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.928, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.04683128398639828, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.04683128398639828, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 353.08, - "cuda_time_us": 349.631, - "pct_cuda_time": 0.7717604002379531, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.143, - "cuda_time_us": 147.424, - "pct_cuda_time": 0.32541738359779315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0031785939357283893, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.984, - "pct_cuda_time": 0.3222387896620647, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.974, - "cuda_time_us": 26.048, - "pct_cuda_time": 0.05749723252628686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.048, - "pct_cuda_time": 0.05749723252628686, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 153.626, - "cuda_time_us": 66.97500000000001, - "pct_cuda_time": 0.1478377283648673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.872, - "pct_cuda_time": 0.02620574111456072, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.76, - "pct_cuda_time": 0.11866750693385987, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.0029644803164466855, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.318, - "cuda_time_us": 109.184, - "pct_cuda_time": 0.24100805574900586, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.001626822035160988, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.447, - "pct_cuda_time": 0.2393812337138449, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.441, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.047678909035925844, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.047678909035925844, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.004, - "cuda_time_us": 1005.502, - "pct_cuda_time": 2.2195017774741435, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.467, - "cuda_time_us": 621.759, - "pct_cuda_time": 1.3724440186698248, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 620.703, - "pct_cuda_time": 1.370113049783624, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.353, - "cuda_time_us": 89.664, - "pct_cuda_time": 0.19792044906468773, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.664, - "pct_cuda_time": 0.19792044906468773, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.288, - "cuda_time_us": 294.079, - "pct_cuda_time": 0.6491373097396314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.079, - "pct_cuda_time": 0.6491373097396314, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 601.768, - "cuda_time_us": 1400.028, - "pct_cuda_time": 3.090361465729129, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.591, - "cuda_time_us": 20.128, - "pct_cuda_time": 0.04442967967940349, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.128, - "pct_cuda_time": 0.04442967967940349, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 434.86, - "cuda_time_us": 350.142, - "pct_cuda_time": 0.7728883596137568, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.959, - "cuda_time_us": 146.4, - "pct_cuda_time": 0.32315705013238627, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0022603334654068548, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.376, - "pct_cuda_time": 0.32089671666697944, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 160.616, - "cuda_time_us": 25.792, - "pct_cuda_time": 0.05693214915993515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.792, - "pct_cuda_time": 0.05693214915993515, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.907, - "cuda_time_us": 67.389, - "pct_cuda_time": 0.1487515741213892, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.327, - "pct_cuda_time": 0.025002731604163518, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.527, - "pct_cuda_time": 0.1203605496760152, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.535, - "pct_cuda_time": 0.003388292841210471, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 50.133, - "cuda_time_us": 110.56099999999999, - "pct_cuda_time": 0.24404758620004616, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.377, - "pct_cuda_time": 0.0030395304510402728, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 109.184, - "pct_cuda_time": 0.24100805574900586, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.866, - "cuda_time_us": 21.856, - "pct_cuda_time": 0.04824399240227756, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.856, - "pct_cuda_time": 0.04824399240227756, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.994, - "cuda_time_us": 1007.902, - "pct_cuda_time": 2.224799434033691, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 38.386, - "cuda_time_us": 623.519, - "pct_cuda_time": 1.3763289668134928, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0022603334654068548, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.495, - "pct_cuda_time": 1.374068633348086, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.87, - "cuda_time_us": 90.016, - "pct_cuda_time": 0.19869743869342132, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 90.016, - "pct_cuda_time": 0.19869743869342132, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.242, - "cuda_time_us": 294.367, - "pct_cuda_time": 0.649773028526777, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.367, - "pct_cuda_time": 0.649773028526777, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 520.283, - "cuda_time_us": 1397.246, - "pct_cuda_time": 3.084220598833854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.991, - "cuda_time_us": 19.392, - "pct_cuda_time": 0.04280506500114231, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.392, - "pct_cuda_time": 0.04280506500114231, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 376.612, - "cuda_time_us": 347.968, - "pct_cuda_time": 0.7680895657135669, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.646, - "cuda_time_us": 145.824, - "pct_cuda_time": 0.32188561255809495, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0016952500990551412, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.056, - "pct_cuda_time": 0.3201903624590398, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.996, - "cuda_time_us": 26.112, - "pct_cuda_time": 0.057638503367874794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.112, - "pct_cuda_time": 0.057638503367874794, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.316, - "cuda_time_us": 66.751, - "pct_cuda_time": 0.14734328041930952, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.025287480644239187, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.792, - "pct_cuda_time": 0.11873814235465385, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.003317657420416506, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 49.904, - "cuda_time_us": 109.281, - "pct_cuda_time": 0.24122216936828758, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0024722397277887474, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.161, - "pct_cuda_time": 0.23874992964049885, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.085, - "cuda_time_us": 21.984, - "pct_cuda_time": 0.04852653408545342, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.984, - "pct_cuda_time": 0.04852653408545342, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.712, - "cuda_time_us": 1007.902, - "pct_cuda_time": 2.224799434033691, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.472, - "cuda_time_us": 623.455, - "pct_cuda_time": 1.376187695971905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0016952500990551412, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.687, - "pct_cuda_time": 1.3744924458728498, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.799, - "cuda_time_us": 89.472, - "pct_cuda_time": 0.19749663653992394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.472, - "pct_cuda_time": 0.19749663653992394, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.647, - "cuda_time_us": 294.975, - "pct_cuda_time": 0.6511151015218622, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.975, - "pct_cuda_time": 0.6511151015218622, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 510.922, - "cuda_time_us": 1395.9969999999998, - "pct_cuda_time": 3.081463610065989, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.741, - "cuda_time_us": 19.936, - "pct_cuda_time": 0.0440058671546397, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.936, - "pct_cuda_time": 0.0440058671546397, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 357.788, - "cuda_time_us": 348.51099999999997, - "pct_cuda_time": 0.7692881605101644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.273, - "cuda_time_us": 146.75199999999998, - "pct_cuda_time": 0.32393403976111984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.016, - "pct_cuda_time": 0.3223094250828587, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.426, - "cuda_time_us": 25.792, - "pct_cuda_time": 0.05693214915993515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.792, - "pct_cuda_time": 0.05693214915993515, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.102, - "cuda_time_us": 66.88, - "pct_cuda_time": 0.1476280294593852, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.025570022327415044, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.984, - "pct_cuda_time": 0.11916195487941762, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.002896052252552533, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.589, - "cuda_time_us": 109.087, - "pct_cuda_time": 0.2407939421297242, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.055, - "pct_cuda_time": 0.0023287615293010075, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.032, - "pct_cuda_time": 0.23846518060042315, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.611, - "cuda_time_us": 21.568, - "pct_cuda_time": 0.04760827361513188, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.568, - "pct_cuda_time": 0.04760827361513188, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.748, - "cuda_time_us": 1005.982, - "pct_cuda_time": 2.220561308786053, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.766, - "cuda_time_us": 622.27, - "pct_cuda_time": 1.3735719780456284, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.0029644803164466855, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 620.927, - "pct_cuda_time": 1.3706074977291818, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.539, - "cuda_time_us": 89.056, - "pct_cuda_time": 0.1965783760696024, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.056, - "pct_cuda_time": 0.1965783760696024, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.446, - "cuda_time_us": 294.656, - "pct_cuda_time": 0.6504109546708224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.656, - "pct_cuda_time": 0.6504109546708224, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 566.56, - "cuda_time_us": 1397.758, - "pct_cuda_time": 3.0853507655665573, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.013, - "cuda_time_us": 21.153, - "pct_cuda_time": 0.04669222050171015, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.153, - "pct_cuda_time": 0.04669222050171015, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 391.984, - "cuda_time_us": 348.192, - "pct_cuda_time": 0.7685840136591245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 37.648, - "cuda_time_us": 147.392, - "pct_cuda_time": 0.32534674817699916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.377, - "pct_cuda_time": 0.0030395304510402728, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.015, - "pct_cuda_time": 0.32230721772595883, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 112.557, - "cuda_time_us": 25.92, - "pct_cuda_time": 0.05721469084311101, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.92, - "pct_cuda_time": 0.05721469084311101, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 165.833, - "cuda_time_us": 66.52799999999999, - "pct_cuda_time": 0.14685103983065156, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.552, - "pct_cuda_time": 0.02549938690662108, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.632, - "pct_cuda_time": 0.11838496525068401, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.491, - "cuda_time_us": 108.352, - "pct_cuda_time": 0.23917153480836284, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.296, - "pct_cuda_time": 0.23684056592216202, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.457, - "cuda_time_us": 21.152, - "pct_cuda_time": 0.04669001314481035, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.152, - "pct_cuda_time": 0.04669001314481035, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 125.003, - "cuda_time_us": 1007.261, - "pct_cuda_time": 2.223384518260912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.754, - "cuda_time_us": 623.358, - "pct_cuda_time": 1.3759735823526231, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.622, - "pct_cuda_time": 1.3743489676743619, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.998, - "cuda_time_us": 89.279, - "pct_cuda_time": 0.1970706166582603, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.279, - "pct_cuda_time": 0.1970706166582603, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.78, - "cuda_time_us": 294.624, - "pct_cuda_time": 0.6503403192500286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.624, - "pct_cuda_time": 0.6503403192500286, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 525.681, - "cuda_time_us": 1396.349, - "pct_cuda_time": 3.082240599694723, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.019, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.04563048183290088, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.04563048183290088, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 379.976, - "cuda_time_us": 350.462, - "pct_cuda_time": 0.7735947138216964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.418, - "cuda_time_us": 147.486, - "pct_cuda_time": 0.3255542397255814, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.023, - "pct_cuda_time": 0.0022581261085070433, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.463, - "pct_cuda_time": 0.3232961136170744, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 110.733, - "cuda_time_us": 25.631, - "pct_cuda_time": 0.056576764699065515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.631, - "pct_cuda_time": 0.056576764699065515, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.482, - "cuda_time_us": 67.32900000000001, - "pct_cuda_time": 0.14861913270740051, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.489, - "pct_cuda_time": 0.02536032342193296, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.336, - "pct_cuda_time": 0.11993894450815122, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0033198647773163177, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.387, - "cuda_time_us": 110.01599999999999, - "pct_cuda_time": 0.24284457668964896, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.672, - "pct_cuda_time": 0.23987788901630244, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.837, - "cuda_time_us": 20.416, - "pct_cuda_time": 0.04506539846654917, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.416, - "pct_cuda_time": 0.04506539846654917, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.839, - "cuda_time_us": 1004.799, - "pct_cuda_time": 2.2179500055735764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.127, - "cuda_time_us": 621.119, - "pct_cuda_time": 1.3710313102539455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0016952500990551412, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 620.351, - "pct_cuda_time": 1.3693360601548905, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.213, - "cuda_time_us": 89.76, - "pct_cuda_time": 0.19813235532706963, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.76, - "pct_cuda_time": 0.19813235532706963, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.251, - "cuda_time_us": 293.92, - "pct_cuda_time": 0.6487863399925613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.92, - "pct_cuda_time": 0.6487863399925613, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 507.643, - "cuda_time_us": 1399.421, - "pct_cuda_time": 3.0890216000909434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.322, - "cuda_time_us": 20.0, - "pct_cuda_time": 0.04414713799622763, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.0, - "pct_cuda_time": 0.04414713799622763, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 366.799, - "cuda_time_us": 350.111, - "pct_cuda_time": 0.7728199315498626, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.821, - "cuda_time_us": 147.263, - "pct_cuda_time": 0.3250619991369235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0022603334654068548, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.239, - "pct_cuda_time": 0.32280166567151664, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.046, - "cuda_time_us": 26.272, - "pct_cuda_time": 0.057991680471844616, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.272, - "pct_cuda_time": 0.057991680471844616, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.72, - "cuda_time_us": 66.912, - "pct_cuda_time": 0.1476986648801792, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.488, - "pct_cuda_time": 0.02535811606503315, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.92, - "pct_cuda_time": 0.1190206840378297, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0033198647773163177, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.035, - "cuda_time_us": 109.664, - "pct_cuda_time": 0.24206758706091536, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.928, - "pct_cuda_time": 0.24044297238265416, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.849, - "cuda_time_us": 22.272, - "pct_cuda_time": 0.04916225287259909, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.272, - "pct_cuda_time": 0.04916225287259909, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 92.905, - "cuda_time_us": 1007.038, - "pct_cuda_time": 2.2228922776722544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.466, - "cuda_time_us": 622.975, - "pct_cuda_time": 1.3751281646599955, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0024722397277887474, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.855, - "pct_cuda_time": 1.3726559249322068, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.252, - "cuda_time_us": 89.919, - "pct_cuda_time": 0.1984833250741396, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.919, - "pct_cuda_time": 0.1984833250741396, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.554, - "cuda_time_us": 294.144, - "pct_cuda_time": 0.649280787938119, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.144, - "pct_cuda_time": 0.649280787938119, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 523.966, - "cuda_time_us": 1398.238, - "pct_cuda_time": 3.0864102968784666, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.11, - "cuda_time_us": 19.871, - "pct_cuda_time": 0.043862388956151965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.871, - "pct_cuda_time": 0.043862388956151965, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 379.203, - "cuda_time_us": 348.543, - "pct_cuda_time": 0.7693587959309584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.44, - "cuda_time_us": 147.2, - "pct_cuda_time": 0.32492293565223535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.0021896980446128905, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.208, - "pct_cuda_time": 0.32273323760762246, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 96.034, - "cuda_time_us": 25.92, - "pct_cuda_time": 0.05721469084311101, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.92, - "pct_cuda_time": 0.05721469084311101, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.34, - "cuda_time_us": 66.14399999999999, - "pct_cuda_time": 0.146003414781124, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.424, - "pct_cuda_time": 0.025216845223445222, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.376, - "pct_cuda_time": 0.11781988188433229, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.232, - "cuda_time_us": 109.27900000000001, - "pct_cuda_time": 0.24121775465448797, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.543, - "pct_cuda_time": 0.2395931399762268, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.266, - "cuda_time_us": 21.665, - "pct_cuda_time": 0.04782238723441358, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.665, - "pct_cuda_time": 0.04782238723441358, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.459, - "cuda_time_us": 1008.159, - "pct_cuda_time": 2.2253667247569426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.943, - "cuda_time_us": 623.328, - "pct_cuda_time": 1.3759073616456288, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.409, - "pct_cuda_time": 0.003110165871834237, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.919, - "pct_cuda_time": 1.3727971957737946, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.976, - "cuda_time_us": 89.696, - "pct_cuda_time": 0.1979910844854817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.696, - "pct_cuda_time": 0.1979910844854817, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.355, - "cuda_time_us": 295.135, - "pct_cuda_time": 0.651468278625832, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.135, - "pct_cuda_time": 0.651468278625832, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 589.6, - "cuda_time_us": 1400.063, - "pct_cuda_time": 3.0904387232206223, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.465, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.04584238809528277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.04584238809528277, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 447.042, - "cuda_time_us": 350.20799999999997, - "pct_cuda_time": 0.7730340451691442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.848, - "cuda_time_us": 148.19199999999998, - "pct_cuda_time": 0.32711263369684823, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0031079585149344255, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.784, - "pct_cuda_time": 0.3240046751819138, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 116.902, - "cuda_time_us": 25.76, - "pct_cuda_time": 0.05686151373914119, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.76, - "pct_cuda_time": 0.05686151373914119, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 219.554, - "cuda_time_us": 67.328, - "pct_cuda_time": 0.1486169253505007, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.552, - "pct_cuda_time": 0.02549938690662108, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.4, - "pct_cuda_time": 0.12008021534973916, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.003037323094140461, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.563, - "cuda_time_us": 108.928, - "pct_cuda_time": 0.24044297238265416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.872, - "pct_cuda_time": 0.23811200349645334, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.387, - "cuda_time_us": 21.12, - "pct_cuda_time": 0.046619377724016385, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.12, - "pct_cuda_time": 0.046619377724016385, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.62, - "cuda_time_us": 1007.967, - "pct_cuda_time": 2.224942912232179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.368, - "cuda_time_us": 623.4879999999999, - "pct_cuda_time": 1.3762605387495985, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.025, - "pct_cuda_time": 0.002262540822306666, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.463, - "pct_cuda_time": 1.3739979979272918, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.973, - "cuda_time_us": 89.408, - "pct_cuda_time": 0.197355365698336, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.408, - "pct_cuda_time": 0.197355365698336, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.431, - "cuda_time_us": 295.071, - "pct_cuda_time": 0.6513270077842442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.071, - "pct_cuda_time": 0.6513270077842442, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 493.277, - "cuda_time_us": 1397.118, - "pct_cuda_time": 3.0839380571506774, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.112, - "cuda_time_us": 20.192, - "pct_cuda_time": 0.04457095052099142, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.192, - "pct_cuda_time": 0.04457095052099142, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 356.554, - "cuda_time_us": 348.8, - "pct_cuda_time": 0.7699260866542099, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.329, - "cuda_time_us": 146.816, - "pct_cuda_time": 0.3240753106027078, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.0021896980446128905, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.824, - "pct_cuda_time": 0.32188561255809495, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.903, - "cuda_time_us": 25.568, - "pct_cuda_time": 0.056437701214377405, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.568, - "pct_cuda_time": 0.056437701214377405, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 149.421, - "cuda_time_us": 67.00800000000001, - "pct_cuda_time": 0.14791057114256106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.424, - "pct_cuda_time": 0.025216845223445222, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.048, - "pct_cuda_time": 0.11930322572100555, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0033905001981102824, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.408, - "cuda_time_us": 109.408, - "pct_cuda_time": 0.24150250369456366, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.003037323094140461, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.032, - "pct_cuda_time": 0.23846518060042315, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.206, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.04563048183290088, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.04563048183290088, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 92.079, - "cuda_time_us": 1007.454, - "pct_cuda_time": 2.2238105381425757, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.28, - "cuda_time_us": 622.8149999999999, - "pct_cuda_time": 1.3747749875560256, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.001626822035160988, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.078, - "pct_cuda_time": 1.3731481655208646, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.518, - "cuda_time_us": 88.928, - "pct_cuda_time": 0.19629583438642656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 88.928, - "pct_cuda_time": 0.19629583438642656, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.397, - "cuda_time_us": 295.711, - "pct_cuda_time": 0.6527397162001235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.711, - "pct_cuda_time": 0.6527397162001235, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 537.204, - "cuda_time_us": 1397.757, - "pct_cuda_time": 3.0853485582096574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.745, - "cuda_time_us": 20.256, - "pct_cuda_time": 0.044712221362579345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.256, - "pct_cuda_time": 0.044712221362579345, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 388.408, - "cuda_time_us": 349.28000000000003, - "pct_cuda_time": 0.7709856179661194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.468, - "cuda_time_us": 146.144, - "pct_cuda_time": 0.3225919667660346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.001626822035160988, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.407, - "pct_cuda_time": 0.32096514473087356, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.306, - "cuda_time_us": 25.952, - "pct_cuda_time": 0.05728532626390497, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.952, - "pct_cuda_time": 0.05728532626390497, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.724, - "cuda_time_us": 67.64800000000001, - "pct_cuda_time": 0.14932327955844035, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.904, - "pct_cuda_time": 0.026276376535354688, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.24, - "pct_cuda_time": 0.11972703824576934, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0033198647773163177, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 46.191, - "cuda_time_us": 109.536, - "pct_cuda_time": 0.2417850453777395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.48, - "pct_cuda_time": 0.2394540764915387, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.895, - "cuda_time_us": 21.568, - "pct_cuda_time": 0.04760827361513188, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.568, - "pct_cuda_time": 0.04760827361513188, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.99, - "cuda_time_us": 1006.653, - "pct_cuda_time": 2.222042445265827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.125, - "cuda_time_us": 621.567, - "pct_cuda_time": 1.372020206145061, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0016952500990551412, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 620.799, - "pct_cuda_time": 1.3703249560460058, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.257, - "cuda_time_us": 89.535, - "pct_cuda_time": 0.19763570002461206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.535, - "pct_cuda_time": 0.19763570002461206, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.014, - "cuda_time_us": 295.551, - "pct_cuda_time": 0.6523865390961536, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.551, - "pct_cuda_time": 0.6523865390961536, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 497.908, - "cuda_time_us": 1395.934, - "pct_cuda_time": 3.081324546581301, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.011, - "cuda_time_us": 20.288, - "pct_cuda_time": 0.04478285678337331, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.288, - "pct_cuda_time": 0.04478285678337331, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 353.743, - "cuda_time_us": 348.64, - "pct_cuda_time": 0.76957290955024, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.862, - "cuda_time_us": 146.112, - "pct_cuda_time": 0.3225213313452406, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0016952500990551412, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.344, - "pct_cuda_time": 0.32082608124618545, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.737, - "cuda_time_us": 25.888, - "pct_cuda_time": 0.05714405542231705, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.888, - "pct_cuda_time": 0.05714405542231705, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 150.227, - "cuda_time_us": 67.073, - "pct_cuda_time": 0.1480540493410488, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.329, - "pct_cuda_time": 0.02500714631796314, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.271, - "pct_cuda_time": 0.11979546630966349, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.473, - "pct_cuda_time": 0.003251436713422165, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.815, - "cuda_time_us": 109.56700000000001, - "pct_cuda_time": 0.24185347344163366, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.087, - "pct_cuda_time": 0.0023993969500949717, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.48, - "pct_cuda_time": 0.2394540764915387, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.093, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.04746700277354395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.04746700277354395, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.394, - "cuda_time_us": 1005.502, - "pct_cuda_time": 2.2195017774741435, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.199, - "cuda_time_us": 622.207, - "pct_cuda_time": 1.3734329145609403, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0031079585149344255, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 620.799, - "pct_cuda_time": 1.3703249560460058, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.109, - "cuda_time_us": 89.504, - "pct_cuda_time": 0.19756727196071788, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.504, - "pct_cuda_time": 0.19756727196071788, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.32, - "cuda_time_us": 293.791, - "pct_cuda_time": 0.6485015909524856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.791, - "pct_cuda_time": 0.6485015909524856, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 526.566, - "cuda_time_us": 1397.723, - "pct_cuda_time": 3.0852735080750633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 21.309, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.04584238809528277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.04584238809528277, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 373.842, - "cuda_time_us": 349.437, - "pct_cuda_time": 0.7713321729993897, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.454, - "cuda_time_us": 147.488, - "pct_cuda_time": 0.32555865443938103, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0022603334654068548, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.464, - "pct_cuda_time": 0.32329832097397415, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.514, - "cuda_time_us": 26.143, - "pct_cuda_time": 0.05770693143176896, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.143, - "pct_cuda_time": 0.05770693143176896, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.227, - "cuda_time_us": 66.87899999999999, - "pct_cuda_time": 0.14762582210248537, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.359, - "pct_cuda_time": 0.02507336702495748, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.209, - "pct_cuda_time": 0.1196586101818752, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.311, - "pct_cuda_time": 0.0028938448956527213, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.516, - "cuda_time_us": 108.92699999999999, - "pct_cuda_time": 0.24044076502575434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.871, - "pct_cuda_time": 0.23810979613955352, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.542, - "cuda_time_us": 21.024, - "pct_cuda_time": 0.046407471461634486, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.024, - "pct_cuda_time": 0.046407471461634486, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.959, - "cuda_time_us": 1006.4939999999999, - "pct_cuda_time": 2.2216914755187567, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.193, - "cuda_time_us": 621.823, - "pct_cuda_time": 1.3725852895114128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.087, - "pct_cuda_time": 1.3709606748331515, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.439, - "cuda_time_us": 89.664, - "pct_cuda_time": 0.19792044906468773, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.664, - "pct_cuda_time": 0.19792044906468773, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.254, - "cuda_time_us": 295.007, - "pct_cuda_time": 0.6511857369426562, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.007, - "pct_cuda_time": 0.6511857369426562, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 519.206, - "cuda_time_us": 1395.392, - "pct_cuda_time": 3.080128159141603, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.106, - "cuda_time_us": 19.808, - "pct_cuda_time": 0.04372332547146385, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.808, - "pct_cuda_time": 0.04372332547146385, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 366.87, - "cuda_time_us": 348.226, - "pct_cuda_time": 0.7686590637937182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.847, - "cuda_time_us": 147.264, - "pct_cuda_time": 0.32506420649382334, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.184, - "pct_cuda_time": 0.002613510569376676, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.08, - "pct_cuda_time": 0.32245069592444664, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 105.585, - "cuda_time_us": 25.408, - "pct_cuda_time": 0.05608452411040758, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.408, - "pct_cuda_time": 0.05608452411040758, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.364, - "cuda_time_us": 66.753, - "pct_cuda_time": 0.14734769513310916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.521, - "pct_cuda_time": 0.02543095884272693, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.696, - "pct_cuda_time": 0.11852623609227193, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0033905001981102824, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.963, - "cuda_time_us": 108.80099999999999, - "pct_cuda_time": 0.2401626380563781, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.001626822035160988, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.064, - "pct_cuda_time": 0.23853581602121715, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.844, - "cuda_time_us": 21.12, - "pct_cuda_time": 0.046619377724016385, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.12, - "pct_cuda_time": 0.046619377724016385, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 105.047, - "cuda_time_us": 1006.238, - "pct_cuda_time": 2.221126392152405, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.226, - "cuda_time_us": 622.943, - "pct_cuda_time": 1.3750575292392013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.887, - "pct_cuda_time": 1.3727265603530003, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 26.714, - "cuda_time_us": 89.504, - "pct_cuda_time": 0.19756727196071788, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.504, - "pct_cuda_time": 0.19756727196071788, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.465, - "cuda_time_us": 293.791, - "pct_cuda_time": 0.6485015909524856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.791, - "pct_cuda_time": 0.6485015909524856, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 522.472, - "cuda_time_us": 1399.645, - "pct_cuda_time": 3.089516048036501, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.012, - "cuda_time_us": 19.776, - "pct_cuda_time": 0.04365269005066988, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.776, - "pct_cuda_time": 0.04365269005066988, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 364.168, - "cuda_time_us": 348.895, - "pct_cuda_time": 0.770135785559692, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.912, - "cuda_time_us": 146.944, - "pct_cuda_time": 0.3243578522858836, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.0021896980446128905, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.952, - "pct_cuda_time": 0.32216815424127077, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.5, - "cuda_time_us": 26.08, - "pct_cuda_time": 0.05756786794708083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.08, - "pct_cuda_time": 0.05756786794708083, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.34, - "cuda_time_us": 67.10499999999999, - "pct_cuda_time": 0.14812468476184273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.552, - "pct_cuda_time": 0.02549938690662108, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.048, - "pct_cuda_time": 0.11930322572100555, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.0033220721342161292, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.064, - "cuda_time_us": 108.766, - "pct_cuda_time": 0.2400853805648847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0016224073213613654, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.031, - "pct_cuda_time": 0.23846297324352336, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.207, - "cuda_time_us": 21.248, - "pct_cuda_time": 0.04690191940719224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.248, - "pct_cuda_time": 0.04690191940719224, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 110.603, - "cuda_time_us": 1009.7260000000001, - "pct_cuda_time": 2.228825653018947, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 47.068, - "cuda_time_us": 624.606, - "pct_cuda_time": 1.3787283637635876, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0024016043069947837, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 623.518, - "pct_cuda_time": 1.376326759456593, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.247, - "cuda_time_us": 89.152, - "pct_cuda_time": 0.19679028233198428, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.152, - "pct_cuda_time": 0.19679028233198428, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.075, - "cuda_time_us": 295.968, - "pct_cuda_time": 0.653307006923375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.968, - "pct_cuda_time": 0.653307006923375, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 626.224, - "cuda_time_us": 1398.301, - "pct_cuda_time": 3.0865493603631546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.945, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.04513603388734313, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.04513603388734313, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 468.148, - "cuda_time_us": 347.96599999999995, - "pct_cuda_time": 0.768085150999767, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.278, - "cuda_time_us": 145.59799999999998, - "pct_cuda_time": 0.3213867498987375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.767, - "pct_cuda_time": 0.0016930427421553295, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 144.831, - "pct_cuda_time": 0.3196937071565822, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.193, - "cuda_time_us": 25.92, - "pct_cuda_time": 0.05721469084311101, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.92, - "pct_cuda_time": 0.05721469084311101, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 249.364, - "cuda_time_us": 67.008, - "pct_cuda_time": 0.14791057114256106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.52, - "pct_cuda_time": 0.025428751485827114, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.144, - "pct_cuda_time": 0.11951513198338744, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 51.298, - "cuda_time_us": 109.44, - "pct_cuda_time": 0.2415731391153576, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.704, - "pct_cuda_time": 0.2399485244370964, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 23.848, - "cuda_time_us": 21.248, - "pct_cuda_time": 0.04690191940719224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.248, - "pct_cuda_time": 0.04690191940719224, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.683, - "cuda_time_us": 1008.6389999999999, - "pct_cuda_time": 2.226426256068852, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.813, - "cuda_time_us": 624.319, - "pct_cuda_time": 1.3780948523333418, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.003037323094140461, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.943, - "pct_cuda_time": 1.3750575292392013, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.574, - "cuda_time_us": 89.312, - "pct_cuda_time": 0.1971434594359541, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.312, - "pct_cuda_time": 0.1971434594359541, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.706, - "cuda_time_us": 295.008, - "pct_cuda_time": 0.6511879442995561, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.008, - "pct_cuda_time": 0.6511879442995561, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 509.106, - "cuda_time_us": 1397.2440000000001, - "pct_cuda_time": 3.0842161841200544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.694, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.0449947630457552, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.0449947630457552, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 368.098, - "cuda_time_us": 348.703, - "pct_cuda_time": 0.7697119730349281, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.352, - "cuda_time_us": 147.04, - "pct_cuda_time": 0.32456975854826553, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.003037323094140461, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.664, - "pct_cuda_time": 0.321532435454125, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.491, - "cuda_time_us": 25.856, - "pct_cuda_time": 0.05707342000152309, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.856, - "pct_cuda_time": 0.05707342000152309, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.501, - "cuda_time_us": 67.071, - "pct_cuda_time": 0.14804963462724918, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.647, - "pct_cuda_time": 0.025709085812103162, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.08, - "pct_cuda_time": 0.11937386114179951, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.067, - "cuda_time_us": 108.736, - "pct_cuda_time": 0.24001915985789038, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.68, - "pct_cuda_time": 0.23768819097168958, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.45, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.04577175267448881, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.04577175267448881, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.802, - "cuda_time_us": 1007.421, - "pct_cuda_time": 2.223737695364882, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.753, - "cuda_time_us": 622.59, - "pct_cuda_time": 1.374278332253568, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0016952500990551412, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.822, - "pct_cuda_time": 1.3725830821545129, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.577, - "cuda_time_us": 89.664, - "pct_cuda_time": 0.19792044906468773, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.664, - "pct_cuda_time": 0.19792044906468773, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.158, - "cuda_time_us": 295.167, - "pct_cuda_time": 0.651538914046626, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.167, - "pct_cuda_time": 0.651538914046626, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 520.405, - "cuda_time_us": 1396.859, - "pct_cuda_time": 3.0833663517136265, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.844, - "cuda_time_us": 19.52, - "pct_cuda_time": 0.043087606684318165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.52, - "pct_cuda_time": 0.043087606684318165, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 377.357, - "cuda_time_us": 349.69399999999996, - "pct_cuda_time": 0.7718994637226412, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 39.522, - "cuda_time_us": 146.207, - "pct_cuda_time": 0.32273103025072264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.184, - "pct_cuda_time": 0.002613510569376676, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.023, - "pct_cuda_time": 0.320117519681346, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.521, - "cuda_time_us": 26.143, - "pct_cuda_time": 0.05770693143176896, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.143, - "pct_cuda_time": 0.05770693143176896, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.508, - "cuda_time_us": 67.04, - "pct_cuda_time": 0.14798120656335503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.648, - "pct_cuda_time": 0.025711293169002968, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.856, - "pct_cuda_time": 0.11887941319624178, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0033905001981102824, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.372, - "cuda_time_us": 110.304, - "pct_cuda_time": 0.24348029547679464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0031785939357283893, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.864, - "pct_cuda_time": 0.24030170154106628, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.403, - "cuda_time_us": 21.728, - "pct_cuda_time": 0.0479614507191017, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.728, - "pct_cuda_time": 0.0479614507191017, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.941, - "cuda_time_us": 1005.9169999999999, - "pct_cuda_time": 2.2204178305875653, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.555, - "cuda_time_us": 622.143, - "pct_cuda_time": 1.3732916437193523, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.407, - "pct_cuda_time": 1.3716670290410913, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.598, - "cuda_time_us": 89.439, - "pct_cuda_time": 0.19742379376223013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.439, - "pct_cuda_time": 0.19742379376223013, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.731, - "cuda_time_us": 294.335, - "pct_cuda_time": 0.6497023931059829, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.335, - "pct_cuda_time": 0.6497023931059829, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 520.6, - "cuda_time_us": 1396.094, - "pct_cuda_time": 3.0816777236852713, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.697, - "cuda_time_us": 19.808, - "pct_cuda_time": 0.04372332547146385, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.808, - "pct_cuda_time": 0.04372332547146385, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 372.643, - "cuda_time_us": 347.775, - "pct_cuda_time": 0.7676635458319031, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.769, - "cuda_time_us": 146.112, - "pct_cuda_time": 0.3225213313452406, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.376, - "pct_cuda_time": 0.32089671666697944, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.828, - "cuda_time_us": 25.536, - "pct_cuda_time": 0.05636706579358344, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.536, - "pct_cuda_time": 0.05636706579358344, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.29, - "cuda_time_us": 66.944, - "pct_cuda_time": 0.14776930030097313, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.025640657748209006, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.824, - "pct_cuda_time": 0.1188087777754478, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0033198647773163177, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.618, - "cuda_time_us": 109.18299999999999, - "pct_cuda_time": 0.24100584839210604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.127, - "pct_cuda_time": 0.23867487950590527, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.952, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.047678909035925844, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.047678909035925844, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.07, - "cuda_time_us": 1006.911, - "pct_cuda_time": 2.222611943345978, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.555, - "cuda_time_us": 622.56, - "pct_cuda_time": 1.3742121115465735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.001626822035160988, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.823, - "pct_cuda_time": 1.3725852895114128, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.276, - "cuda_time_us": 89.312, - "pct_cuda_time": 0.1971434594359541, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.312, - "pct_cuda_time": 0.1971434594359541, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.972, - "cuda_time_us": 295.039, - "pct_cuda_time": 0.6512563723634502, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.039, - "pct_cuda_time": 0.6512563723634502, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 522.097, - "cuda_time_us": 1395.904, - "pct_cuda_time": 3.0812583258743067, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.983, - "cuda_time_us": 20.193, - "pct_cuda_time": 0.044573157877891234, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.193, - "pct_cuda_time": 0.044573157877891234, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 374.526, - "cuda_time_us": 348.576, - "pct_cuda_time": 0.7694316387086522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.587, - "cuda_time_us": 147.136, - "pct_cuda_time": 0.32478166481064746, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.4, - "pct_cuda_time": 0.32315705013238627, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.437, - "cuda_time_us": 25.664, - "pct_cuda_time": 0.0566496074767593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.664, - "pct_cuda_time": 0.0566496074767593, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.56, - "cuda_time_us": 66.88000000000001, - "pct_cuda_time": 0.14762802945938522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.36, - "pct_cuda_time": 0.025075574381857296, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.016, - "pct_cuda_time": 0.11923259030021158, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0033198647773163177, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.328, - "cuda_time_us": 108.896, - "pct_cuda_time": 0.24037233696186022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.216, - "pct_cuda_time": 0.0026841459901706397, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.68, - "pct_cuda_time": 0.23768819097168958, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.984, - "cuda_time_us": 21.951, - "pct_cuda_time": 0.04845369130775964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.951, - "pct_cuda_time": 0.04845369130775964, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.192, - "cuda_time_us": 1005.184, - "pct_cuda_time": 2.2187998379800034, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.696, - "cuda_time_us": 622.5279999999999, - "pct_cuda_time": 1.3741414761257795, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.377, - "pct_cuda_time": 0.0030395304510402728, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.151, - "pct_cuda_time": 1.3711019456747393, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.851, - "cuda_time_us": 89.152, - "pct_cuda_time": 0.19679028233198428, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.152, - "pct_cuda_time": 0.19679028233198428, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.015, - "cuda_time_us": 293.504, - "pct_cuda_time": 0.6478680795222398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.504, - "pct_cuda_time": 0.6478680795222398, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 541.895, - "cuda_time_us": 1396.605, - "pct_cuda_time": 3.082805683061075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.774, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.04555984641210691, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.04555984641210691, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 387.292, - "cuda_time_us": 348.671, - "pct_cuda_time": 0.7696413376141342, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.036, - "cuda_time_us": 147.647, - "pct_cuda_time": 0.32590962418645103, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.023, - "pct_cuda_time": 0.0022581261085070433, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.624, - "pct_cuda_time": 0.323651498077944, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.613, - "cuda_time_us": 25.504, - "pct_cuda_time": 0.05629643037278948, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.504, - "pct_cuda_time": 0.05629643037278948, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 177.728, - "cuda_time_us": 66.592, - "pct_cuda_time": 0.14699231067223953, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.392, - "pct_cuda_time": 0.025146209802651257, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.856, - "pct_cuda_time": 0.11887941319624178, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 46.74, - "cuda_time_us": 108.928, - "pct_cuda_time": 0.24044297238265416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0031079585149344255, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.52, - "pct_cuda_time": 0.23733501386771974, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.47, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.04725509651116206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.04725509651116206, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.734, - "cuda_time_us": 1005.886, - "pct_cuda_time": 2.2203494025236714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.598, - "cuda_time_us": 623.231, - "pct_cuda_time": 1.375693248026347, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.001626822035160988, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.494, - "pct_cuda_time": 1.3740664259911863, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.768, - "cuda_time_us": 88.992, - "pct_cuda_time": 0.1964371052280145, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 88.992, - "pct_cuda_time": 0.1964371052280145, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.384, - "cuda_time_us": 293.663, - "pct_cuda_time": 0.6482190492693097, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.663, - "pct_cuda_time": 0.6482190492693097, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 570.203, - "cuda_time_us": 1398.7150000000001, - "pct_cuda_time": 3.087463206119677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.516, - "cuda_time_us": 19.968, - "pct_cuda_time": 0.04407650257543367, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.968, - "pct_cuda_time": 0.04407650257543367, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 419.308, - "cuda_time_us": 349.982, - "pct_cuda_time": 0.772535182509787, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.451, - "cuda_time_us": 147.391, - "pct_cuda_time": 0.32534454082009934, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.002966687673346497, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.047, - "pct_cuda_time": 0.3223778531467528, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 123.318, - "cuda_time_us": 26.144, - "pct_cuda_time": 0.057709138788668755, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.144, - "pct_cuda_time": 0.057709138788668755, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 168.508, - "cuda_time_us": 67.52000000000001, - "pct_cuda_time": 0.1490407378752645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.808, - "pct_cuda_time": 0.026064470272972796, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.176, - "pct_cuda_time": 0.11958576740418142, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0033905001981102824, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.811, - "cuda_time_us": 108.927, - "pct_cuda_time": 0.2404407650257544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.191, - "pct_cuda_time": 0.2388161503474932, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.756, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.047678909035925844, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.047678909035925844, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.32, - "cuda_time_us": 1007.165, - "pct_cuda_time": 2.22317261199853, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.011, - "cuda_time_us": 622.302, - "pct_cuda_time": 1.3736426134664224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0024016043069947837, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.214, - "pct_cuda_time": 1.3712410091594276, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.621, - "cuda_time_us": 89.632, - "pct_cuda_time": 0.19784981364389378, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.632, - "pct_cuda_time": 0.19784981364389378, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.598, - "cuda_time_us": 295.231, - "pct_cuda_time": 0.651680184888214, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.231, - "pct_cuda_time": 0.651680184888214, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 562.766, - "cuda_time_us": 1397.215, - "pct_cuda_time": 3.0841521707699595, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.494, - "cuda_time_us": 19.68, - "pct_cuda_time": 0.04344078378828799, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.68, - "pct_cuda_time": 0.04344078378828799, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 412.692, - "cuda_time_us": 348.545, - "pct_cuda_time": 0.7693632106447581, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.403, - "cuda_time_us": 146.336, - "pct_cuda_time": 0.3230157792907984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.002119062623818926, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.376, - "pct_cuda_time": 0.32089671666697944, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.674, - "cuda_time_us": 26.144, - "pct_cuda_time": 0.057709138788668755, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.144, - "pct_cuda_time": 0.057709138788668755, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 208.775, - "cuda_time_us": 66.753, - "pct_cuda_time": 0.14734769513310916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.025287480644239187, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.792, - "pct_cuda_time": 0.11873814235465385, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.0033220721342161292, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.298, - "cuda_time_us": 109.312, - "pct_cuda_time": 0.2412905974321817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.576, - "pct_cuda_time": 0.23966598275392054, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.359, - "cuda_time_us": 21.888, - "pct_cuda_time": 0.04831462782307152, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.888, - "pct_cuda_time": 0.04831462782307152, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.044, - "cuda_time_us": 1007.102, - "pct_cuda_time": 2.223033548513842, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 39.149, - "cuda_time_us": 623.039, - "pct_cuda_time": 1.3752694355015833, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.983, - "pct_cuda_time": 1.3729384666153823, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.419, - "cuda_time_us": 89.408, - "pct_cuda_time": 0.197355365698336, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.408, - "pct_cuda_time": 0.197355365698336, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.202, - "cuda_time_us": 294.655, - "pct_cuda_time": 0.6504087473139226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.655, - "pct_cuda_time": 0.6504087473139226, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 581.988, - "cuda_time_us": 1398.268, - "pct_cuda_time": 3.086476517585461, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.667, - "cuda_time_us": 20.0, - "pct_cuda_time": 0.04414713799622763, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.0, - "pct_cuda_time": 0.04414713799622763, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 432.165, - "cuda_time_us": 348.094, - "pct_cuda_time": 0.768367692682943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.295, - "cuda_time_us": 146.59199999999998, - "pct_cuda_time": 0.32358086265715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.856, - "pct_cuda_time": 0.3219562479788889, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 161.782, - "cuda_time_us": 26.207, - "pct_cuda_time": 0.05784820227335688, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 26.207, - "pct_cuda_time": 0.05784820227335688, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.684, - "cuda_time_us": 66.399, - "pct_cuda_time": 0.14656629079057593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.025287480644239187, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.567, - "pct_cuda_time": 0.11824148705219628, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.003037323094140461, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 46.926, - "cuda_time_us": 108.896, - "pct_cuda_time": 0.24037233696186022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.84, - "pct_cuda_time": 0.2380413680756594, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.634, - "cuda_time_us": 21.92, - "pct_cuda_time": 0.04838526324386549, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.92, - "pct_cuda_time": 0.04838526324386549, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.481, - "cuda_time_us": 1008.2539999999999, - "pct_cuda_time": 2.2255764236624245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.791, - "cuda_time_us": 623.007, - "pct_cuda_time": 1.3751988000807893, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0022603334654068548, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.983, - "pct_cuda_time": 1.3729384666153823, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.257, - "cuda_time_us": 89.664, - "pct_cuda_time": 0.19792044906468773, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.664, - "pct_cuda_time": 0.19792044906468773, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.16, - "cuda_time_us": 295.583, - "pct_cuda_time": 0.6524571745169476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.583, - "pct_cuda_time": 0.6524571745169476, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 526.639, - "cuda_time_us": 1396.4470000000001, - "pct_cuda_time": 3.0824569206709045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.629, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.04527730472893106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.04527730472893106, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 383.569, - "cuda_time_us": 348.993, - "pct_cuda_time": 0.7703521065358735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.239, - "cuda_time_us": 147.648, - "pct_cuda_time": 0.32591183154335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0031079585149344255, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 146.24, - "pct_cuda_time": 0.32280387302841645, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 113.72, - "cuda_time_us": 25.856, - "pct_cuda_time": 0.05707342000152309, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.856, - "pct_cuda_time": 0.05707342000152309, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.813, - "cuda_time_us": 66.72, - "pct_cuda_time": 0.14727485235541538, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.295, - "pct_cuda_time": 0.024932096183369556, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.08, - "pct_cuda_time": 0.11937386114179951, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.345, - "pct_cuda_time": 0.002968895030246308, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.404, - "cuda_time_us": 108.769, - "pct_cuda_time": 0.2400920026355842, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.025, - "pct_cuda_time": 0.002262540822306666, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 107.744, - "pct_cuda_time": 0.2378294618132775, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.96, - "cuda_time_us": 21.44, - "pct_cuda_time": 0.04732573193195602, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.44, - "pct_cuda_time": 0.04732573193195602, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.835, - "cuda_time_us": 1005.5020000000001, - "pct_cuda_time": 2.219501777474144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.16, - "cuda_time_us": 622.2710000000001, - "pct_cuda_time": 1.3735741854025283, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0016952500990551412, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.503, - "pct_cuda_time": 1.3718789353034733, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.631, - "cuda_time_us": 89.28, - "pct_cuda_time": 0.19707282401516013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.28, - "pct_cuda_time": 0.19707282401516013, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.935, - "cuda_time_us": 293.951, - "pct_cuda_time": 0.6488547680564555, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.951, - "pct_cuda_time": 0.6488547680564555, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 527.522, - "cuda_time_us": 1399.134, - "pct_cuda_time": 3.0883880886606976, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.721, - "cuda_time_us": 19.648, - "pct_cuda_time": 0.043370148367494026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.648, - "pct_cuda_time": 0.043370148367494026, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 383.383, - "cuda_time_us": 350.20799999999997, - "pct_cuda_time": 0.7730340451691442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.634, - "cuda_time_us": 146.84799999999998, - "pct_cuda_time": 0.3241459460235017, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.0021896980446128905, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.856, - "pct_cuda_time": 0.3219562479788889, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.201, - "cuda_time_us": 25.984, - "pct_cuda_time": 0.05735596168469895, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.984, - "pct_cuda_time": 0.05735596168469895, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 173.094, - "cuda_time_us": 67.26299999999999, - "pct_cuda_time": 0.14847344715201294, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.025640657748209006, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 54.144, - "pct_cuda_time": 0.11951513198338744, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.003317657420416506, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.947, - "cuda_time_us": 110.113, - "pct_cuda_time": 0.24305869030893068, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.665, - "pct_cuda_time": 0.0036752492381859504, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.448, - "pct_cuda_time": 0.2393834410707447, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.425, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.046478106882428455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.046478106882428455, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.127, - "cuda_time_us": 1008.222, - "pct_cuda_time": 2.2255057882416307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.487, - "cuda_time_us": 624.574, - "pct_cuda_time": 1.3786577283427937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 623.838, - "pct_cuda_time": 1.3770331136645326, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.484, - "cuda_time_us": 88.896, - "pct_cuda_time": 0.19622519896563256, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 88.896, - "pct_cuda_time": 0.19622519896563256, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.962, - "cuda_time_us": 294.752, - "pct_cuda_time": 0.6506228609332043, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.752, - "pct_cuda_time": 0.6506228609332043, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 513.795, - "cuda_time_us": 1395.101, - "pct_cuda_time": 3.079485818283758, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.441, - "cuda_time_us": 19.392, - "pct_cuda_time": 0.04280506500114231, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 19.392, - "pct_cuda_time": 0.04280506500114231, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 373.287, - "cuda_time_us": 348.032, - "pct_cuda_time": 0.7682308365551547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.289, - "cuda_time_us": 146.368, - "pct_cuda_time": 0.3230864147115923, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.632, - "pct_cuda_time": 0.32146180003333114, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 105.766, - "cuda_time_us": 25.792, - "pct_cuda_time": 0.05693214915993515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.792, - "pct_cuda_time": 0.05693214915993515, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.898, - "cuda_time_us": 66.624, - "pct_cuda_time": 0.14706294609303347, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.36, - "pct_cuda_time": 0.025075574381857296, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.728, - "pct_cuda_time": 0.1185968715130659, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0033905001981102824, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.585, - "cuda_time_us": 109.248, - "pct_cuda_time": 0.24114932659059382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0016246146782611767, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.512, - "pct_cuda_time": 0.23952471191233263, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.709, - "cuda_time_us": 22.4, - "pct_cuda_time": 0.04944479455577495, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.4, - "pct_cuda_time": 0.04944479455577495, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.606, - "cuda_time_us": 1005.277, - "pct_cuda_time": 2.219005122171686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.747, - "cuda_time_us": 622.3660000000001, - "pct_cuda_time": 1.3737838843080104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.248, - "pct_cuda_time": 0.0027547814109646043, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.118, - "pct_cuda_time": 1.3710291028970458, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.488, - "cuda_time_us": 89.535, - "pct_cuda_time": 0.19763570002461206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.535, - "pct_cuda_time": 0.19763570002461206, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.075, - "cuda_time_us": 293.376, - "pct_cuda_time": 0.6475855378390638, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.376, - "pct_cuda_time": 0.6475855378390638, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 507.404, - "cuda_time_us": 1397.981, - "pct_cuda_time": 3.085843006155215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 20.925, - "cuda_time_us": 20.32, - "pct_cuda_time": 0.044853492204167275, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 20.32, - "pct_cuda_time": 0.044853492204167275, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 349.08, - "cuda_time_us": 347.903, - "pct_cuda_time": 0.7679460875150791, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.737, - "cuda_time_us": 146.591, - "pct_cuda_time": 0.3235786553002502, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.023, - "pct_cuda_time": 0.0022581261085070433, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.568, - "pct_cuda_time": 0.32132052919174325, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2048, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.044, - "cuda_time_us": 25.472, - "pct_cuda_time": 0.05622579495199551, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 25.472, - "pct_cuda_time": 0.05622579495199551, - "trace": "_C::rotary_embedding(int64[2048], bfloat16[2048, 4096], bfloat16[2048, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 148.836, - "cuda_time_us": 66.56, - "pct_cuda_time": 0.14692167525144556, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 11.328, - "pct_cuda_time": 0.025004938961063327, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2048], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 53.92, - "pct_cuda_time": 0.1190206840378297, - "trace": "_vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.002896052252552533, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], None, None, bfloat16[2048, 32, 128], int32[3], int32[3], None, None, None, 1024, 1024, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2048, 32, 128], bfloat16[2048, 8, 128], bfloat16[2048, 8, 128], bfloat16[2048, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.357, - "cuda_time_us": 109.28, - "pct_cuda_time": 0.24121996201138776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0016952500990551412, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 108.512, - "pct_cuda_time": 0.23952471191233263, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2048, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 22.892, - "cuda_time_us": 22.24, - "pct_cuda_time": 0.049091617451805125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.24, - "pct_cuda_time": 0.049091617451805125, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.905, - "cuda_time_us": 1007.5179999999999, - "pct_cuda_time": 2.2239518089841632, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.051, - "cuda_time_us": 623.39, - "pct_cuda_time": 1.3760442177734171, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.0036024064604921746, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 621.758, - "pct_cuda_time": 1.372441811312925, - "trace": "mm(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2048, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2048, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.117, - "cuda_time_us": 89.824, - "pct_cuda_time": 0.19827362616865754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 89.824, - "pct_cuda_time": 0.19827362616865754, - "trace": "_C::silu_and_mul(bfloat16[2048, 14336], bfloat16[2048, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.608, - "cuda_time_us": 294.304, - "pct_cuda_time": 0.6496339650420888, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.304, - "pct_cuda_time": 0.6496339650420888, - "trace": "mm(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2048, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2048, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.584, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.046478106882428455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.046478106882428455, - "trace": "_C::fused_add_rms_norm(bfloat16[2048, 4096], bfloat16[2048, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 92.449, - "cuda_time_us": 357.375, - "pct_cuda_time": 0.7888541720700925, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.008, - "pct_cuda_time": 0.006639729554632635, - "trace": "index_select(bfloat16[2048, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.002330968886200819, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 353.311, - "pct_cuda_time": 0.7798834736292589, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 28717.346, - "cuda_time_us": 135.809, - "pct_cuda_time": 0.2997789332064839, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.007204812920984349, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.004873844034783531, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.004944479455577495, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.004803208613989567, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.004873844034783531, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.004873844034783531, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.004803208613989567, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.009818323490361026, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 5.088, - "pct_cuda_time": 0.01123103190624031, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 37.664, - "pct_cuda_time": 0.08313789027449588, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.008, - "pct_cuda_time": 0.06844572274935132, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.004803208613989567, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 5.184, - "pct_cuda_time": 0.011442938168622203, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 30.656, - "pct_cuda_time": 0.0676687331206177, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 3.105, - "pct_cuda_time": 0.00685384317391434, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - }, - "decode_1": { - "metadata": { - "num_running_seqs": 2 - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 6546.937000000003, - "pct_cuda_time": 93.25662770134517, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 6539.961000000002, - "pct_cuda_time": 93.15725936545853, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 210.0510000000001, - "pct_cuda_time": 2.992032442850031, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.416, - "pct_cuda_time": 0.0629028915245618, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 205.63500000000008, - "pct_cuda_time": 2.929129551325469, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 2072.865, - "pct_cuda_time": 29.526540362332604, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 669.5989999999998, - "pct_cuda_time": 9.53797854663837, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 669.5989999999998, - "pct_cuda_time": 9.53797854663837, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 119.64800000000002, - "pct_cuda_time": 1.7043037058720043, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 119.64800000000002, - "pct_cuda_time": 1.7043037058720043, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 718.3059999999998, - "pct_cuda_time": 10.231776358569265, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 74.52799999999999, - "pct_cuda_time": 1.0616002489906118, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cuda_time_us": 519.138, - "pct_cuda_time": 7.394764787200627, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cuda_time_us": 79.00899999999999, - "pct_cuda_time": 1.125429020938429, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 45.631, - "pct_cuda_time": 0.6499823014396013, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 565.312, - "pct_cuda_time": 8.052481751252962, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cuda_time_us": 497.53299999999996, - "pct_cuda_time": 7.087016378824684, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cuda_time_us": 67.779, - "pct_cuda_time": 0.9654653724282777, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 4257.044999999999, - "pct_cuda_time": 60.63868656027584, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 2577.4979999999996, - "pct_cuda_time": 36.7146913720052, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2577.4979999999996, - "pct_cuda_time": 36.7146913720052, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 283.93299999999994, - "pct_cuda_time": 4.044430864864901, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 283.93299999999994, - "pct_cuda_time": 4.044430864864901, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 1395.6140000000005, - "pct_cuda_time": 19.879564323405756, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1395.6140000000005, - "pct_cuda_time": 19.879564323405756, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 348.031, - "pct_cuda_time": 4.957462916708507, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.072, - "pct_cuda_time": 0.04375853323447778, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010483815254093634, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 344.223, - "pct_cuda_time": 4.903220568219936, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 125.377, - "pct_cuda_time": 1.785909381946328, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 16.160000000000004, - "pct_cuda_time": 0.23018811753553417, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.16, - "pct_cuda_time": 0.059256347088355324, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.672, - "pct_cuda_time": 0.06654943596076827, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 34.144, - "pct_cuda_time": 0.48635786417903937, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.84, - "pct_cuda_time": 0.39656170743745484, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 1.952, - "pct_cuda_time": 0.027804901326074417, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.928, - "pct_cuda_time": 0.07019598039697476, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 28.736, - "pct_cuda_time": 0.40932461296417755, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.785, - "pct_cuda_time": 0.03967041505794942, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 17527.309, - "cuda_time_us": 6546.937000000003, - "pct_cuda_time": 93.25662770134517, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 62.404, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[2]) <- embedding(bfloat16[128256, 4096], int64[2], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 959.598, - "cuda_time_us": 207.712, - "pct_cuda_time": 2.9587149919270335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 57.168, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.0629028915245618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.0629028915245618, - "trace": "_C::rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 685.478, - "cuda_time_us": 67.29599999999999, - "pct_cuda_time": 0.9585853686677785, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 92.389, - "cuda_time_us": 23.199, - "pct_cuda_time": 0.33045384521700844, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 23.199, - "pct_cuda_time": 0.33045384521700844, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 206.993, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05469816654309721, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05469816654309721, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 262.18, - "cuda_time_us": 22.401, - "pct_cuda_time": 0.3190868824822711, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03281889992585833, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.417, - "pct_cuda_time": 0.23384890628594454, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.368, - "pct_cuda_time": 0.03373053603490995, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.11, - "cuda_time_us": 17.856, - "pct_cuda_time": 0.2543464744254021, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.743, - "pct_cuda_time": 0.22424823851249467, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.0300982359129074, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.803, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 159.805, - "cuda_time_us": 132.768, - "pct_cuda_time": 1.8911891082275862, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 64.696, - "cuda_time_us": 80.704, - "pct_cuda_time": 1.149573133514093, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.704, - "pct_cuda_time": 1.149573133514093, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 34.326, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12671741915817522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12671741915817522, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 42.938, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6148985555553179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6148985555553179, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 645.513, - "cuda_time_us": 204.19299999999998, - "pct_cuda_time": 2.908589250243398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 20.193, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 460.848, - "cuda_time_us": 64.416, - "pct_cuda_time": 0.9175617437604557, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 38.342, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29400264516914754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29400264516914754, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 133.694, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05469816654309721, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05469816654309721, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 205.784, - "cuda_time_us": 22.432, - "pct_cuda_time": 0.3195284562225929, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.035553808253013186, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.16, - "pct_cuda_time": 0.23018811753553414, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.83, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.24933247582561813, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.2197043022814405, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.029628173544177662, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 20.238, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 122.673, - "cuda_time_us": 133.281, - "pct_cuda_time": 1.8984964414142034, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 42.047, - "cuda_time_us": 81.184, - "pct_cuda_time": 1.1564104043319803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.184, - "pct_cuda_time": 1.1564104043319803, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 25.25, - "cuda_time_us": 8.961, - "pct_cuda_time": 0.12764329958143078, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.961, - "pct_cuda_time": 0.12764329958143078, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 37.206, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.6144427375007921, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.6144427375007921, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 617.093, - "cuda_time_us": 204.92700000000002, - "pct_cuda_time": 2.9190445768690845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.077, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 445.254, - "cuda_time_us": 65.055, - "pct_cuda_time": 0.9266638605367682, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.252, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2962817354417766, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2962817354417766, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 120.207, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05469816654309721, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05469816654309721, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 210.486, - "cuda_time_us": 22.592, - "pct_cuda_time": 0.3218075464952219, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.035553808253013186, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.23109975364458574, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03418635408943576, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.020967630508187268, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.433, - "cuda_time_us": 17.823, - "pct_cuda_time": 0.25387641205667233, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.712, - "pct_cuda_time": 0.2238066647721728, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.030069747284499543, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.456, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 114.179, - "cuda_time_us": 133.408, - "pct_cuda_time": 1.9003054693181025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 41.906, - "cuda_time_us": 81.344, - "pct_cuda_time": 1.1586894946046093, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.344, - "pct_cuda_time": 1.1586894946046093, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 24.273, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.12899650943080426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.12899650943080426, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 34.588, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.6126194652826888, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.6126194652826888, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 556.35, - "cuda_time_us": 203.77800000000002, - "pct_cuda_time": 2.902677859848767, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.248, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04831671377973588, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04831671377973588, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 396.163, - "cuda_time_us": 64.86600000000001, - "pct_cuda_time": 0.9239716851522253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.408, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.3035748243141896, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.3035748243141896, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 108.957, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.052419076270468164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.052419076270468164, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 181.405, - "cuda_time_us": 22.402, - "pct_cuda_time": 0.319101126796475, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.129, - "pct_cuda_time": 0.22974654379521228, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.497, - "pct_cuda_time": 0.035568052567217116, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.728, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.24887665777109236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.21879266617238885, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.996, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 108.114, - "cuda_time_us": 132.38400000000001, - "pct_cuda_time": 1.885719291573277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 38.155, - "cuda_time_us": 80.704, - "pct_cuda_time": 1.149573133514093, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.704, - "pct_cuda_time": 1.149573133514093, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 23.244, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 34.029, - "cuda_time_us": 42.944, - "pct_cuda_time": 0.6117078291736373, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.944, - "pct_cuda_time": 0.6117078291736373, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 503.126, - "cuda_time_us": 202.882, - "pct_cuda_time": 2.889914954322044, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.92, - "cuda_time_us": 3.553, - "pct_cuda_time": 0.050610048366568856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.553, - "pct_cuda_time": 0.050610048366568856, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 363.112, - "cuda_time_us": 64.001, - "pct_cuda_time": 0.9116503533658242, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 35.192, - "cuda_time_us": 20.577, - "pct_cuda_time": 0.2931052533742999, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.577, - "pct_cuda_time": 0.2931052533742999, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.353, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.05150744016141655, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.05150744016141655, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.879, - "cuda_time_us": 22.368, - "pct_cuda_time": 0.31861682011354125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03327471798038414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.192, - "pct_cuda_time": 0.23064393559005994, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.036009626307539, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.11, - "cuda_time_us": 17.439999999999998, - "pct_cuda_time": 0.24842083971656648, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.21879266617238885, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.029628173544177662, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.981, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.45, - "cuda_time_us": 132.16, - "pct_cuda_time": 1.882528565191596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.299, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.1418242265871543, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.1418242265871543, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.467, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12398251083102037, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12398251083102037, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.021, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.6167218277734211, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.6167218277734211, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 560.181, - "cuda_time_us": 203.745, - "pct_cuda_time": 2.9022077974800373, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.698, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.04833095809393981, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.04833095809393981, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 390.268, - "cuda_time_us": 64.865, - "pct_cuda_time": 0.923957440838021, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.788, - "cuda_time_us": 21.377, - "pct_cuda_time": 0.3045007047374451, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.377, - "pct_cuda_time": 0.3045007047374451, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.072, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 177.989, - "cuda_time_us": 22.176, - "pct_cuda_time": 0.3158819117863864, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03281889992585833, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.128, - "pct_cuda_time": 0.22973229948100835, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.253, - "cuda_time_us": 17.664, - "pct_cuda_time": 0.2516115660982472, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.584, - "pct_cuda_time": 0.22198339255406954, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.029628173544177662, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.403, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 109.379, - "cuda_time_us": 132.255, - "pct_cuda_time": 1.8838817750409693, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.736, - "cuda_time_us": 79.743, - "pct_cuda_time": 1.135884347564115, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.743, - "pct_cuda_time": 1.135884347564115, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.918, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12671741915817522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12671741915817522, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.382, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.6212800083186792, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.6212800083186792, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 492.78, - "cuda_time_us": 204.416, - "pct_cuda_time": 2.9117657323108754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.082, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 352.152, - "cuda_time_us": 64.512, - "pct_cuda_time": 0.9189291979240334, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.82, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29400264516914754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29400264516914754, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.48, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.0537865304340456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.0537865304340456, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.853, - "cuda_time_us": 22.432, - "pct_cuda_time": 0.3195284562225929, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.192, - "pct_cuda_time": 0.23064393559005994, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.525, - "cuda_time_us": 17.664, - "pct_cuda_time": 0.2516115660982472, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.22107175644501795, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03053980965322928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.497, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.134, - "cuda_time_us": 133.344, - "pct_cuda_time": 1.8993938332090505, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.676, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1436474988052576, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1436474988052576, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 19.999, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.13264305386701075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.13264305386701075, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.582, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.6231032805367824, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.6231032805367824, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 483.808, - "cuda_time_us": 204.67100000000002, - "pct_cuda_time": 2.9153980324328783, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.581, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 336.466, - "cuda_time_us": 64.671, - "pct_cuda_time": 0.9211940438824584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.146, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.29810500765987985, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.29810500765987985, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 93.512, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 153.618, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.3231750006587994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.192, - "pct_cuda_time": 0.23064393559005994, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.72, - "pct_cuda_time": 0.038744534634693864, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.948, - "cuda_time_us": 17.343, - "pct_cuda_time": 0.24703914123878515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.232, - "pct_cuda_time": 0.2169693939542856, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.030069747284499543, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.033, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.107, - "cuda_time_us": 133.536, - "pct_cuda_time": 1.9021287415362058, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.157, - "cuda_time_us": 80.639, - "pct_cuda_time": 1.1486472530908376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.639, - "pct_cuda_time": 1.1486472530908376, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.313, - "cuda_time_us": 8.929, - "pct_cuda_time": 0.12718748152690496, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.929, - "pct_cuda_time": 0.12718748152690496, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 38.578, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6262940069184632, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6262940069184632, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 483.148, - "cuda_time_us": 205.05700000000002, - "pct_cuda_time": 2.920896337715596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.964, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 340.131, - "cuda_time_us": 64.993, - "pct_cuda_time": 0.9257807130561242, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.119, - "cuda_time_us": 20.992, - "pct_cuda_time": 0.2990166437689315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.992, - "pct_cuda_time": 0.2990166437689315, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 92.328, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05105162210689074, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05105162210689074, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.707, - "cuda_time_us": 22.496, - "pct_cuda_time": 0.3204400923316445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.384, - "pct_cuda_time": 0.2333788439172148, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.020055994399135645, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.985, - "cuda_time_us": 17.921, - "pct_cuda_time": 0.2552723548486576, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.809, - "pct_cuda_time": 0.22518836324995414, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.426, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.087, - "cuda_time_us": 133.6, - "pct_cuda_time": 1.9030403776452574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.762, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.1459265890778865, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.1459265890778865, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.362, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12717323721270105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12717323721270105, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.716, - "cuda_time_us": 44.224, - "pct_cuda_time": 0.6299405513546695, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.224, - "pct_cuda_time": 0.6299405513546695, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 526.842, - "cuda_time_us": 202.912, - "pct_cuda_time": 2.8903422837481623, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.253, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 377.604, - "cuda_time_us": 64.512, - "pct_cuda_time": 0.9189291979240334, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.652, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.2921793729510443, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.2921793729510443, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.749, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05333071237951979, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05333071237951979, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.354, - "cuda_time_us": 22.432, - "pct_cuda_time": 0.3195284562225929, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.23109975364458574, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03418635408943576, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.019600176344609834, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.892, - "cuda_time_us": 17.823999999999998, - "pct_cuda_time": 0.25389065637087627, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.744, - "pct_cuda_time": 0.22426248282669858, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.029628173544177662, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.297, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.768, - "cuda_time_us": 131.744, - "pct_cuda_time": 1.8766029304827605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.69, - "cuda_time_us": 79.872, - "pct_cuda_time": 1.137721864096422, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.872, - "pct_cuda_time": 1.137721864096422, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 28.176, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12717323721270105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12717323721270105, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.585, - "cuda_time_us": 42.944, - "pct_cuda_time": 0.6117078291736373, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.944, - "pct_cuda_time": 0.6117078291736373, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 478.074, - "cuda_time_us": 204.31900000000002, - "pct_cuda_time": 2.910384033833094, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.132, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 341.208, - "cuda_time_us": 64.959, - "pct_cuda_time": 0.9252964063731908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.366, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.295370099332725, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.295370099332725, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.027, - "cuda_time_us": 4.096, - "pct_cuda_time": 0.0583447109793037, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.096, - "pct_cuda_time": 0.0583447109793037, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.35, - "cuda_time_us": 22.461999999999996, - "pct_cuda_time": 0.3199557856487108, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.223, - "pct_cuda_time": 0.2310855093303818, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.021409204248509145, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.077, - "cuda_time_us": 17.665, - "pct_cuda_time": 0.2516258104124511, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.552, - "pct_cuda_time": 0.22152757449954374, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.0300982359129074, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.546, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 92.583, - "cuda_time_us": 132.832, - "pct_cuda_time": 1.8921007443366378, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.392, - "cuda_time_us": 80.32, - "pct_cuda_time": 1.1441033168597834, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.32, - "pct_cuda_time": 1.1441033168597834, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.124, - "cuda_time_us": 9.376, - "pct_cuda_time": 0.13355468997606235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.376, - "pct_cuda_time": 0.13355468997606235, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.158, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.6144427375007921, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.6144427375007921, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 546.643, - "cuda_time_us": 204.098, - "pct_cuda_time": 2.907236040394025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.95, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 382.154, - "cuda_time_us": 64.449, - "pct_cuda_time": 0.9180318061291856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.874, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2962817354417766, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2962817354417766, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.357, - "cuda_time_us": 3.681, - "pct_cuda_time": 0.052433320584672094, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.681, - "pct_cuda_time": 0.052433320584672094, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.211, - "cuda_time_us": 22.464, - "pct_cuda_time": 0.31998427427711873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03281889992585833, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.128, - "pct_cuda_time": 0.22973229948100835, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.035553808253013186, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.02187926661723889, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.315, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.24933247582561813, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.392, - "pct_cuda_time": 0.21924848422691465, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.805, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 114.706, - "cuda_time_us": 133.089, - "pct_cuda_time": 1.8957615330870485, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 46.825, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1518522237867221, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1518522237867221, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 23.113, - "cuda_time_us": 8.897, - "pct_cuda_time": 0.12673166347237916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.897, - "pct_cuda_time": 0.12673166347237916, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.961, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.617177645827947, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.617177645827947, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 585.511, - "cuda_time_us": 203.806, - "pct_cuda_time": 2.9030767006464773, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.75, - "cuda_time_us": 3.391, - "pct_cuda_time": 0.04830246946553195, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.391, - "pct_cuda_time": 0.04830246946553195, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 429.721, - "cuda_time_us": 64.47999999999999, - "pct_cuda_time": 0.9184733798695073, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.229, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29400264516914754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29400264516914754, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.683, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05469816654309721, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05469816654309721, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 220.799, - "cuda_time_us": 22.272, - "pct_cuda_time": 0.31724936594996384, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03327471798038414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.192, - "pct_cuda_time": 0.23064393559005994, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.789, - "cuda_time_us": 17.728, - "pct_cuda_time": 0.25252320220729885, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.22243921060859534, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.91, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.044214351289003584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.044214351289003584, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 105.955, - "cuda_time_us": 132.83100000000002, - "pct_cuda_time": 1.8920865000224343, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 38.975, - "cuda_time_us": 80.768, - "pct_cuda_time": 1.1504847696231448, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.768, - "pct_cuda_time": 1.1504847696231448, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.913, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.016, - "cuda_time_us": 43.327, - "pct_cuda_time": 0.617163401513743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.327, - "pct_cuda_time": 0.617163401513743, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 511.812, - "cuda_time_us": 205.086, - "pct_cuda_time": 2.9213094228275094, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.689, - "cuda_time_us": 3.361, - "pct_cuda_time": 0.047875140039414, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.361, - "pct_cuda_time": 0.047875140039414, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 366.299, - "cuda_time_us": 64.767, - "pct_cuda_time": 0.9225614980460357, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.856, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.3035748243141896, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.3035748243141896, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.036, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.052419076270468164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.052419076270468164, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 171.596, - "cuda_time_us": 22.304, - "pct_cuda_time": 0.31770518400448966, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.16, - "pct_cuda_time": 0.23018811753553414, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03418635408943576, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.971, - "cuda_time_us": 17.471, - "pct_cuda_time": 0.24886241345688842, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.391, - "pct_cuda_time": 0.21923423991271077, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.029628173544177662, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.795, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.168, - "cuda_time_us": 133.79000000000002, - "pct_cuda_time": 1.9057467973440045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.239, - "cuda_time_us": 80.831, - "pct_cuda_time": 1.1513821614179927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.831, - "pct_cuda_time": 1.1513821614179927, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.85, - "cuda_time_us": 8.831, - "pct_cuda_time": 0.12579153873491966, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.831, - "pct_cuda_time": 0.12579153873491966, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.945, - "cuda_time_us": 44.128, - "pct_cuda_time": 0.6285730971910922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.128, - "pct_cuda_time": 0.6285730971910922, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 501.279, - "cuda_time_us": 205.406, - "pct_cuda_time": 2.9258676033727675, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.494, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 362.033, - "cuda_time_us": 64.768, - "pct_cuda_time": 0.9225757423602398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.337, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.2958259173872508, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.2958259173872508, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 105.387, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05333071237951979, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05333071237951979, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.499, - "cuda_time_us": 22.464, - "pct_cuda_time": 0.31998427427711873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03327471798038414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.064, - "pct_cuda_time": 0.2288206633719567, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.02279090272629051, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.907, - "cuda_time_us": 17.791999999999998, - "pct_cuda_time": 0.2534348383163504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.22107175644501795, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.751, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.713, - "cuda_time_us": 134.206, - "pct_cuda_time": 1.9116724320528398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.362, - "cuda_time_us": 80.831, - "pct_cuda_time": 1.1513821614179927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.831, - "pct_cuda_time": 1.1513821614179927, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.392, - "cuda_time_us": 8.799, - "pct_cuda_time": 0.12533572068039384, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.799, - "pct_cuda_time": 0.12533572068039384, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.1, - "cuda_time_us": 44.576, - "pct_cuda_time": 0.6349545499544536, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.576, - "pct_cuda_time": 0.6349545499544536, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 487.046, - "cuda_time_us": 204.127, - "pct_cuda_time": 2.907649125505939, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.691, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 349.54, - "cuda_time_us": 64.384, - "pct_cuda_time": 0.91710592570593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.535, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.29354682711462177, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.29354682711462177, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.107, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 162.883, - "cuda_time_us": 22.528, - "pct_cuda_time": 0.3208959103861703, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03327471798038414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.23109975364458574, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.293, - "cuda_time_us": 17.6, - "pct_cuda_time": 0.2506999299891956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.456, - "pct_cuda_time": 0.2201601203359663, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03053980965322928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.311, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.849, - "cuda_time_us": 133.151, - "pct_cuda_time": 1.8966446805676922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.887, - "cuda_time_us": 80.767, - "pct_cuda_time": 1.1504705253089407, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.767, - "pct_cuda_time": 1.1504705253089407, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.481, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.882, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6217358263732051, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6217358263732051, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 485.966, - "cuda_time_us": 205.95100000000002, - "pct_cuda_time": 2.9336307546139104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.321, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 343.019, - "cuda_time_us": 64.959, - "pct_cuda_time": 0.9252964063731908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.071, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.29719337155082826, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.29719337155082826, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.145, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.055609802652148835, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.055609802652148835, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.119, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.32226336454974774, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.288, - "pct_cuda_time": 0.2320113897536374, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.020967630508187268, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.612, - "cuda_time_us": 17.567, - "pct_cuda_time": 0.25022986762046584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.455, - "pct_cuda_time": 0.22014587602176236, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.181, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.422, - "cuda_time_us": 134.464, - "pct_cuda_time": 1.9153474651174542, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.13, - "cuda_time_us": 81.025, - "pct_cuda_time": 1.1541455583735554, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.025, - "pct_cuda_time": 1.1541455583735554, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 19.611, - "cuda_time_us": 8.671, - "pct_cuda_time": 0.1235124484622906, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.671, - "pct_cuda_time": 0.1235124484622906, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.557, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.6376894582816084, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.6376894582816084, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 490.083, - "cuda_time_us": 203.137, - "pct_cuda_time": 2.893547254444047, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.833, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 351.38, - "cuda_time_us": 63.870999999999995, - "pct_cuda_time": 0.9097985925193129, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.118, - "cuda_time_us": 20.544, - "pct_cuda_time": 0.2926351910055701, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.544, - "pct_cuda_time": 0.2926351910055701, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.297, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.163, - "cuda_time_us": 22.208, - "pct_cuda_time": 0.31633772984091224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03281889992585833, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.128, - "pct_cuda_time": 0.22973229948100835, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019144358290084026, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.452, - "cuda_time_us": 17.471, - "pct_cuda_time": 0.24886241345688842, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.263, - "pct_cuda_time": 0.21741096769460752, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.0314514457622809, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.059, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04650768587583656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04650768587583656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.73, - "cuda_time_us": 132.673, - "pct_cuda_time": 1.8898358983782129, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.551, - "cuda_time_us": 79.937, - "pct_cuda_time": 1.1386477445196777, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.937, - "pct_cuda_time": 1.1386477445196777, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.535, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.12489414694007199, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.12489414694007199, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.192, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6262940069184632, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6262940069184632, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 471.608, - "cuda_time_us": 205.02499999999998, - "pct_cuda_time": 2.9204405196610694, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.49, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 337.577, - "cuda_time_us": 65.12100000000001, - "pct_cuda_time": 0.9276039852742277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.695, - "cuda_time_us": 20.544, - "pct_cuda_time": 0.2926351910055701, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.544, - "pct_cuda_time": 0.2926351910055701, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 94.969, - "cuda_time_us": 3.745, - "pct_cuda_time": 0.05334495669372372, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.745, - "pct_cuda_time": 0.05334495669372372, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 153.933, - "cuda_time_us": 22.272, - "pct_cuda_time": 0.31724936594996384, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.032, - "pct_cuda_time": 0.2283648453174309, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 32.768, - "cuda_time_us": 18.560000000000002, - "pct_cuda_time": 0.2643744716249699, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.448, - "pct_cuda_time": 0.2342904800262664, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.457, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 91.544, - "cuda_time_us": 133.504, - "pct_cuda_time": 1.9016729234816798, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.76, - "cuda_time_us": 81.856, - "pct_cuda_time": 1.165982583477022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.856, - "pct_cuda_time": 1.165982583477022, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 19.54, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.049, - "cuda_time_us": 42.912, - "pct_cuda_time": 0.6112520111191114, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.912, - "pct_cuda_time": 0.6112520111191114, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 586.731, - "cuda_time_us": 204.416, - "pct_cuda_time": 2.9117657323108754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.786, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 437.128, - "cuda_time_us": 65.024, - "pct_cuda_time": 0.9262222867964462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.115, - "cuda_time_us": 21.281, - "pct_cuda_time": 0.30313325057386764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.281, - "pct_cuda_time": 0.30313325057386764, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 96.304, - "cuda_time_us": 3.711, - "pct_cuda_time": 0.052860650010790045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.711, - "pct_cuda_time": 0.052860650010790045, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 244.558, - "cuda_time_us": 22.464, - "pct_cuda_time": 0.31998427427711873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.23109975364458574, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.022, - "cuda_time_us": 17.567999999999998, - "pct_cuda_time": 0.2502441119346697, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.456, - "pct_cuda_time": 0.2201601203359663, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.16, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.77, - "cuda_time_us": 132.89600000000002, - "pct_cuda_time": 1.8930123804456898, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.229, - "cuda_time_us": 80.672, - "pct_cuda_time": 1.1491173154595673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.672, - "pct_cuda_time": 1.1491173154595673, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.274, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.022, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.6194567361005759, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.6194567361005759, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 527.03, - "cuda_time_us": 204.767, - "pct_cuda_time": 2.916765486596455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.82, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 380.082, - "cuda_time_us": 65.312, - "pct_cuda_time": 0.9303246492871784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.2, - "cuda_time_us": 21.856, - "pct_cuda_time": 0.31132373124112833, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.856, - "pct_cuda_time": 0.31132373124112833, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.146, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.05150744016141655, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.05150744016141655, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 174.871, - "cuda_time_us": 22.208, - "pct_cuda_time": 0.31633772984091224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.192, - "pct_cuda_time": 0.23064393559005994, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.271, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.25115574804372137, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.22107175644501795, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.77, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.804, - "cuda_time_us": 132.895, - "pct_cuda_time": 1.8929981361314858, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.68, - "cuda_time_us": 80.191, - "pct_cuda_time": 1.1422658003274764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.191, - "pct_cuda_time": 1.1422658003274764, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.134, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.963, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6262940069184632, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6262940069184632, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 513.332, - "cuda_time_us": 204.416, - "pct_cuda_time": 2.9117657323108754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.703, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 374.738, - "cuda_time_us": 64.576, - "pct_cuda_time": 0.9198408340330848, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.95, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.29491428127819913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.29491428127819913, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.452, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 171.563, - "cuda_time_us": 22.72, - "pct_cuda_time": 0.32363081871332516, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.036921262416590626, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.384, - "pct_cuda_time": 0.2333788439172148, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03464217214396157, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.97, - "cuda_time_us": 17.439999999999998, - "pct_cuda_time": 0.24842083971656648, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.21879266617238885, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.029628173544177662, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.385, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.609, - "cuda_time_us": 133.344, - "pct_cuda_time": 1.8993938332090505, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.323, - "cuda_time_us": 80.0, - "pct_cuda_time": 1.1395451363145255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.0, - "pct_cuda_time": 1.1395451363145255, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.58, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12717323721270105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12717323721270105, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.554, - "cuda_time_us": 44.416, - "pct_cuda_time": 0.6326754596818245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.416, - "pct_cuda_time": 0.6326754596818245, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 504.027, - "cuda_time_us": 204.00000000000003, - "pct_cuda_time": 2.90584009760204, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.82, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 358.781, - "cuda_time_us": 64.546, - "pct_cuda_time": 0.919413504606967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.319, - "cuda_time_us": 20.513, - "pct_cuda_time": 0.2921936172652483, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.513, - "pct_cuda_time": 0.2921936172652483, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.576, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.481, - "cuda_time_us": 22.464000000000002, - "pct_cuda_time": 0.31998427427711873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03190726381680671, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.256, - "pct_cuda_time": 0.23155557169911156, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.836, - "cuda_time_us": 17.857, - "pct_cuda_time": 0.254360718739606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.712, - "pct_cuda_time": 0.2238066647721728, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.145, - "pct_cuda_time": 0.03055405396743321, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.327, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.044214351289003584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.044214351289003584, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.77, - "cuda_time_us": 133.02200000000002, - "pct_cuda_time": 1.8948071640353854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.186, - "cuda_time_us": 79.903, - "pct_cuda_time": 1.1381634378367442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.903, - "pct_cuda_time": 1.1381634378367442, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.149, - "cuda_time_us": 9.439, - "pct_cuda_time": 0.13445208177091006, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.439, - "pct_cuda_time": 0.13445208177091006, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.408, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6221916444277309, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6221916444277309, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 507.758, - "cuda_time_us": 205.05800000000002, - "pct_cuda_time": 2.9209105820297996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.689, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04558180545258102, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04558180545258102, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 365.365, - "cuda_time_us": 64.866, - "pct_cuda_time": 0.923971685152225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.806, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.30220737015061216, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.30220737015061216, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 107.586, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.0537865304340456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.0537865304340456, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 162.888, - "cuda_time_us": 22.304, - "pct_cuda_time": 0.31770518400448966, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.064, - "pct_cuda_time": 0.2288206633719567, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.11, - "cuda_time_us": 17.57, - "pct_cuda_time": 0.25027260056307765, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.457, - "pct_cuda_time": 0.22017436465017023, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.0300982359129074, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.755, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.504, - "cuda_time_us": 133.824, - "pct_cuda_time": 1.9062311040269384, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.526, - "cuda_time_us": 81.216, - "pct_cuda_time": 1.156866222386506, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.216, - "pct_cuda_time": 1.156866222386506, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.946, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.12808487332175267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.12808487332175267, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.389, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.6212800083186792, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.6212800083186792, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 482.415, - "cuda_time_us": 204.80200000000002, - "pct_cuda_time": 2.917264037593593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.122, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.04877253183426169, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.04877253183426169, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 340.196, - "cuda_time_us": 65.282, - "pct_cuda_time": 0.9298973198610606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.106, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.30311900625966376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.30311900625966376, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 92.952, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05333071237951979, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05333071237951979, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.321, - "cuda_time_us": 22.529, - "pct_cuda_time": 0.32091015470037426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.449, - "pct_cuda_time": 0.23430472434047034, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019144358290084026, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.735, - "cuda_time_us": 17.729, - "pct_cuda_time": 0.25253744652150273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.617, - "pct_cuda_time": 0.22245345492279928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.513, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.561, - "cuda_time_us": 132.928, - "pct_cuda_time": 1.8934681985002155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.679, - "cuda_time_us": 80.576, - "pct_cuda_time": 1.14774986129599, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.576, - "pct_cuda_time": 1.14774986129599, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.375, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.12352669277649457, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.12352669277649457, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.92, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6221916444277309, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6221916444277309, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 493.447, - "cuda_time_us": 203.358, - "pct_cuda_time": 2.8966952478831156, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.12, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.047860895725210066, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 353.992, - "cuda_time_us": 64.19, - "pct_cuda_time": 0.9143425287503673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.496, - "cuda_time_us": 20.607, - "pct_cuda_time": 0.2935325828004178, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.607, - "pct_cuda_time": 0.2935325828004178, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 110.169, - "cuda_time_us": 3.679, - "pct_cuda_time": 0.05240483195626423, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.679, - "pct_cuda_time": 0.05240483195626423, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 153.538, - "cuda_time_us": 22.176, - "pct_cuda_time": 0.3158819117863864, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.192, - "pct_cuda_time": 0.23064393559005994, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03418635408943576, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 32.924, - "cuda_time_us": 17.728, - "pct_cuda_time": 0.25252320220729885, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.648, - "pct_cuda_time": 0.22289502866312116, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.029628173544177662, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.008, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.482, - "cuda_time_us": 132.64, - "pct_cuda_time": 1.889365836009483, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.209, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.1418242265871543, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.1418242265871543, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.489, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12626160110364942, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12626160110364942, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.565, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.6212800083186792, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.6212800083186792, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 533.664, - "cuda_time_us": 204.255, - "pct_cuda_time": 2.909472397724042, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.367, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04650768587583656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04650768587583656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 392.841, - "cuda_time_us": 64.89500000000001, - "pct_cuda_time": 0.9243847702641392, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.186, - "cuda_time_us": 20.575, - "pct_cuda_time": 0.29307676474589195, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.575, - "pct_cuda_time": 0.29307676474589195, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 94.801, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 206.555, - "cuda_time_us": 22.752, - "pct_cuda_time": 0.324086636767851, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03327471798038414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.32, - "pct_cuda_time": 0.23246720780816318, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.624, - "pct_cuda_time": 0.03737708047111644, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.020967630508187268, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.89, - "cuda_time_us": 17.856, - "pct_cuda_time": 0.2543464744254021, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.712, - "pct_cuda_time": 0.2238066647721728, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03053980965322928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.648, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.935, - "cuda_time_us": 132.831, - "pct_cuda_time": 1.892086500022434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.325, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1472940432414642, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1472940432414642, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.13, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12945232748533006, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12945232748533006, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.891, - "cuda_time_us": 43.199, - "pct_cuda_time": 0.6153401292956397, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.199, - "pct_cuda_time": 0.6153401292956397, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 519.008, - "cuda_time_us": 202.399, - "pct_cuda_time": 2.8830349505615454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.503, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.04833095809393981, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.04833095809393981, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 377.068, - "cuda_time_us": 64.256, - "pct_cuda_time": 0.9152826534878268, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 39.756, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.29445846322367336, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.29445846322367336, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.921, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.65, - "cuda_time_us": 22.496, - "pct_cuda_time": 0.3204400923316445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032363081871332515, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.256, - "pct_cuda_time": 0.23155557169911156, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.297, - "cuda_time_us": 17.375999999999998, - "pct_cuda_time": 0.24750920360751488, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.232, - "pct_cuda_time": 0.2169693939542856, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03053980965322928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.369, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.044670169343529395, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.754, - "cuda_time_us": 131.614, - "pct_cuda_time": 1.8747511696362493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.386, - "cuda_time_us": 80.192, - "pct_cuda_time": 1.14228004464168, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.192, - "pct_cuda_time": 1.14228004464168, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.119, - "cuda_time_us": 8.671, - "pct_cuda_time": 0.1235124484622906, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.671, - "pct_cuda_time": 0.1235124484622906, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.938, - "cuda_time_us": 42.751, - "pct_cuda_time": 0.6089586765322783, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.751, - "pct_cuda_time": 0.6089586765322783, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 491.397, - "cuda_time_us": 203.935, - "pct_cuda_time": 2.904914217178784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.951, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 347.45, - "cuda_time_us": 65.151, - "pct_cuda_time": 0.9280313147003455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.937, - "cuda_time_us": 20.991, - "pct_cuda_time": 0.2990023994547275, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.991, - "pct_cuda_time": 0.2990023994547275, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.48, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05196325821594235, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.808, - "cuda_time_us": 22.592, - "pct_cuda_time": 0.3218075464952219, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03281889992585833, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.448, - "pct_cuda_time": 0.2342904800262664, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.036009626307539, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.878, - "cuda_time_us": 17.92, - "pct_cuda_time": 0.2552581105344537, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.808, - "pct_cuda_time": 0.2251741189357502, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.608, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04649344156163263, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.615, - "cuda_time_us": 132.192, - "pct_cuda_time": 1.8829843832461217, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.229, - "cuda_time_us": 80.351, - "pct_cuda_time": 1.1445448906001054, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.351, - "pct_cuda_time": 1.1445448906001054, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.237, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12443832888554618, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.227, - "cuda_time_us": 43.105, - "pct_cuda_time": 0.6140011637604702, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.105, - "pct_cuda_time": 0.6140011637604702, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 540.564, - "cuda_time_us": 203.773, - "pct_cuda_time": 2.902606638277747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 22.348, - "cuda_time_us": 3.423, - "pct_cuda_time": 0.04875828752005776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.423, - "pct_cuda_time": 0.04875828752005776, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 392.609, - "cuda_time_us": 64.41499999999999, - "pct_cuda_time": 0.9175474994462517, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.618, - "cuda_time_us": 20.415, - "pct_cuda_time": 0.29079767447326293, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.415, - "pct_cuda_time": 0.29079767447326293, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.5, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 177.54, - "cuda_time_us": 22.592, - "pct_cuda_time": 0.3218075464952219, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03281889992585833, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.384, - "pct_cuda_time": 0.2333788439172148, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.036921262416590626, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.01868854023555822, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 55.509, - "cuda_time_us": 17.695999999999998, - "pct_cuda_time": 0.25206738415277297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.584, - "pct_cuda_time": 0.22198339255406954, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.771, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.842, - "cuda_time_us": 132.767, - "pct_cuda_time": 1.8911748639133825, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.128, - "cuda_time_us": 80.256, - "pct_cuda_time": 1.1431916807507319, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.256, - "pct_cuda_time": 1.1431916807507319, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.656, - "cuda_time_us": 8.639, - "pct_cuda_time": 0.12305663040776481, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.639, - "pct_cuda_time": 0.12305663040776481, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.315, - "cuda_time_us": 43.872, - "pct_cuda_time": 0.6249265527548857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.872, - "pct_cuda_time": 0.6249265527548857, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 499.387, - "cuda_time_us": 205.85399999999998, - "pct_cuda_time": 2.9322490561361287, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.686, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.047846651411006136, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.047846651411006136, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 349.877, - "cuda_time_us": 65.119, - "pct_cuda_time": 0.9275754966458197, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.886, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.297649189605354, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.297649189605354, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.091, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.052874894324993975, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.8, - "cuda_time_us": 22.752, - "pct_cuda_time": 0.324086636767851, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03327471798038414, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.32, - "pct_cuda_time": 0.23246720780816318, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.023246720780816316, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.84, - "cuda_time_us": 17.759, - "pct_cuda_time": 0.2529647759476207, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.647, - "pct_cuda_time": 0.22288078434891723, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030083991598703473, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.532, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.045125987398055206, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.625, - "cuda_time_us": 134.208, - "pct_cuda_time": 1.9117009206812479, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.169, - "cuda_time_us": 81.088, - "pct_cuda_time": 1.155042950168403, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.088, - "pct_cuda_time": 1.155042950168403, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.911, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.12261505666744295, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.12261505666744295, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.835, - "cuda_time_us": 44.512, - "pct_cuda_time": 0.6340429138454019, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.512, - "pct_cuda_time": 0.6340429138454019, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 495.971, - "cuda_time_us": 203.67999999999998, - "pct_cuda_time": 2.901281917056781, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.782, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.046949259616158444, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 352.949, - "cuda_time_us": 64.288, - "pct_cuda_time": 0.9157384715423526, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.335, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2962817354417766, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.2962817354417766, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.156, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.056065620706674646, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.056065620706674646, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.719, - "cuda_time_us": 22.24, - "pct_cuda_time": 0.316793547895438, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03281889992585833, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.968, - "pct_cuda_time": 0.22745320920837925, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03509799019848738, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021423448562713075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 65], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.114, - "cuda_time_us": 17.312, - "pct_cuda_time": 0.2465975674984633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.199, - "pct_cuda_time": 0.21649933158555587, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.0300982359129074, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.232, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04603762350710683, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.377, - "cuda_time_us": 132.86399999999998, - "pct_cuda_time": 1.8925565623911633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.33, - "cuda_time_us": 80.064, - "pct_cuda_time": 1.140456772423577, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.064, - "pct_cuda_time": 1.140456772423577, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.253, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12717323721270105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12717323721270105, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.124, - "cuda_time_us": 43.872, - "pct_cuda_time": 0.6249265527548857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.872, - "pct_cuda_time": 0.6249265527548857, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.743, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04740507767068425, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 89.715, - "cuda_time_us": 348.031, - "pct_cuda_time": 4.957462916708507, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.072, - "pct_cuda_time": 0.04375853323447778, - "trace": "index_select(bfloat16[2, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010483815254093634, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 344.223, - "pct_cuda_time": 4.903220568219936, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 702.081, - "cuda_time_us": 125.377, - "pct_cuda_time": 1.785909381946328, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.036009626307539, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.624, - "pct_cuda_time": 0.03737708047111644, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03190726381680671, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.0314514457622809, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.0314514457622809, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.030995627707755092, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.030995627707755092, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.16, - "pct_cuda_time": 0.059256347088355324, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.672, - "pct_cuda_time": 0.06654943596076827, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.144, - "pct_cuda_time": 0.48635786417903937, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.84, - "pct_cuda_time": 0.39656170743745484, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 1.952, - "pct_cuda_time": 0.027804901326074417, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.928, - "pct_cuda_time": 0.07019598039697476, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 28.736, - "pct_cuda_time": 0.40932461296417755, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.785, - "pct_cuda_time": 0.03967041505794942, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - } -} \ No newline at end of file diff --git a/H100_llama8b_pp1_tp1/profiling_bs2_pl128.json b/H100_llama8b_pp1_tp1/profiling_bs2_pl128.json deleted file mode 100644 index 9efe8cfa9d7bc43252a5e13c62c7b4974a2ed028..0000000000000000000000000000000000000000 --- a/H100_llama8b_pp1_tp1/profiling_bs2_pl128.json +++ /dev/null @@ -1,17561 +0,0 @@ -{ - "context": { - "python_version": "3.12.9 | packaged by Anaconda, Inc. | (main, Feb 6 2025, 18:56:27) [GCC 11.2.0]", - "torch_version": "2.5.1+cu124", - "engine_args": { - "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "served_model_name": null, - "tokenizer": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "task": "auto", - "skip_tokenizer_init": false, - "tokenizer_mode": "auto", - "trust_remote_code": false, - "allowed_local_media_path": null, - "download_dir": null, - "load_format": "dummy", - "config_format": "auto", - "dtype": "auto", - "kv_cache_dtype": "auto", - "seed": 0, - "max_model_len": null, - "distributed_executor_backend": null, - "pipeline_parallel_size": 1, - "tensor_parallel_size": 1, - "max_parallel_loading_workers": null, - "block_size": null, - "enable_prefix_caching": false, - "disable_sliding_window": false, - "use_v2_block_manager": true, - "swap_space": 4, - "cpu_offload_gb": 0, - "gpu_memory_utilization": 0.9, - "max_num_batched_tokens": 8000, - "max_num_partial_prefills": 1, - "max_long_partial_prefills": 1, - "long_prefill_token_threshold": 0, - "max_num_seqs": 256, - "max_logprobs": 20, - "disable_log_stats": false, - "revision": null, - "code_revision": null, - "rope_scaling": null, - "rope_theta": null, - "hf_overrides": null, - "tokenizer_revision": null, - "quantization": null, - "enforce_eager": true, - "max_seq_len_to_capture": 8192, - "disable_custom_all_reduce": false, - "tokenizer_pool_size": 0, - "tokenizer_pool_type": "ray", - "tokenizer_pool_extra_config": null, - "limit_mm_per_prompt": null, - "mm_processor_kwargs": null, - "disable_mm_preprocessor_cache": false, - "enable_lora": false, - "enable_lora_bias": false, - "max_loras": 1, - "max_lora_rank": 16, - "enable_prompt_adapter": false, - "max_prompt_adapters": 1, - "max_prompt_adapter_token": 0, - "fully_sharded_loras": false, - "lora_extra_vocab_size": 256, - "long_lora_scaling_factors": null, - "lora_dtype": "auto", - "max_cpu_loras": null, - "device": "auto", - "num_scheduler_steps": 1, - "multi_step_stream_outputs": true, - "ray_workers_use_nsight": false, - "num_gpu_blocks_override": null, - "num_lookahead_slots": 0, - "model_loader_extra_config": null, - "ignore_patterns": [], - "preemption_mode": null, - "scheduler_delay_factor": 0.0, - "enable_chunked_prefill": null, - "guided_decoding_backend": "xgrammar", - "logits_processor_pattern": null, - "speculative_model": null, - "speculative_model_quantization": null, - "speculative_draft_tensor_parallel_size": null, - "num_speculative_tokens": null, - "speculative_disable_mqa_scorer": false, - "speculative_max_model_len": null, - "speculative_disable_by_batch_size": null, - "ngram_prompt_lookup_max": null, - "ngram_prompt_lookup_min": null, - "spec_decoding_acceptance_method": "rejection_sampler", - "typical_acceptance_sampler_posterior_threshold": null, - "typical_acceptance_sampler_posterior_alpha": null, - "qlora_adapter_name_or_path": null, - "disable_logprobs_during_spec_decoding": null, - "otlp_traces_endpoint": null, - "collect_detailed_traces": null, - "disable_async_output_proc": false, - "scheduling_policy": "fcfs", - "scheduler_cls": "vllm.core.scheduler.Scheduler", - "override_neuron_config": null, - "override_pooler_config": null, - "compilation_config": null, - "worker_cls": "auto", - "kv_transfer_config": null, - "generation_config": null, - "override_generation_config": null, - "enable_sleep_mode": false, - "model_impl": "auto", - "calculate_kv_scales": false, - "additional_config": null - }, - "prompt_len": 0, - "batch_size": 2, - "num_steps": 2, - "complete_num_requests_per_step": null, - "save_chrome_traces_folder": null - }, - "prefill": { - "metadata": { - "num_running_seqs": null - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 8538.874, - "pct_cuda_time": 94.81778038949706, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 9.728, - "pct_cuda_time": 0.1080221312118, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cuda_time_us": 9.728, - "pct_cuda_time": 0.1080221312118, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 8524.666000000001, - "pct_cuda_time": 94.6600112241746, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 287.5809999999999, - "pct_cuda_time": 3.19337094120278, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 282.9729999999999, - "pct_cuda_time": 3.1422025632603483, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 2197.7960000000003, - "pct_cuda_time": 24.404873343829074, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 897.3770000000001, - "pct_cuda_time": 9.964697372579302, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 897.3770000000001, - "pct_cuda_time": 9.964697372579302, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 148.705, - "pct_cuda_time": 1.6512573007659046, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 148.705, - "pct_cuda_time": 1.6512573007659046, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 460.7059999999999, - "pct_cuda_time": 5.115793994866727, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 94.17700000000002, - "pct_cuda_time": 1.045764828447131, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 320.89900000000006, - "pct_cuda_time": 3.5633422989037227, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 45.63, - "pct_cuda_time": 0.5066868675158752, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 691.008, - "pct_cuda_time": 7.673124675617136, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 691.008, - "pct_cuda_time": 7.673124675617136, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 6039.288999999999, - "pct_cuda_time": 67.06176693914271, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 3338.5590000000007, - "pct_cuda_time": 37.07218938695887, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 23.745000000000008, - "pct_cuda_time": 0.2636703850353816, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 3314.813999999999, - "pct_cuda_time": 36.808519001923464, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 378.62100000000004, - "pct_cuda_time": 4.2043017415237385, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 378.62100000000004, - "pct_cuda_time": 4.2043017415237385, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 2322.109, - "pct_cuda_time": 25.785275810660124, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2322.109, - "pct_cuda_time": 25.785275810660124, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 349.632, - "pct_cuda_time": 3.882400676381996, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.136, - "pct_cuda_time": 0.03482292387748816, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 345.76, - "pct_cuda_time": 3.839405025472036, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 117.05600000000001, - "pct_cuda_time": 1.2998189341209356, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 10.048000000000002, - "pct_cuda_time": 0.11157549079113555, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.192, - "pct_cuda_time": 0.0465490104892954, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 33.728, - "pct_cuda_time": 0.3745240996619645, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.808, - "pct_cuda_time": 0.3087869474442573, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 1.92, - "pct_cuda_time": 0.02132015747601316, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.736, - "pct_cuda_time": 0.05258972177416578, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 27.456, - "pct_cuda_time": 0.30487825190698814, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.592, - "pct_cuda_time": 0.028782212592617765, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 31018.4, - "cuda_time_us": 8538.874, - "pct_cuda_time": 94.81778038949706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 120.109, - "cuda_time_us": 9.728, - "pct_cuda_time": 0.1080221312118, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 9.728, - "pct_cuda_time": 0.1080221312118, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[256]) <- embedding(bfloat16[128256, 4096], int64[256], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1479.239, - "cuda_time_us": 267.841, - "pct_cuda_time": 2.974173072152521, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 84.435, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "_C::rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 1098.727, - "cuda_time_us": 70.624, - "pct_cuda_time": 0.7842264591593506, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 149.566, - "cuda_time_us": 29.44, - "pct_cuda_time": 0.3269090812988684, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 29.44, - "pct_cuda_time": 0.3269090812988684, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 354.716, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 397.215, - "cuda_time_us": 14.655999999999999, - "pct_cuda_time": 0.16274386873356708, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.04, - "pct_cuda_time": 0.0337569160036875, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.113, - "pct_cuda_time": 0.11229726695568805, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.016689685774191547, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 105.412, - "cuda_time_us": 21.952, - "pct_cuda_time": 0.24376046714241714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.952, - "pct_cuda_time": 0.24376046714241714, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 42.889, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 209.088, - "cuda_time_us": 188.19299999999998, - "pct_cuda_time": 2.089741872855908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 75.269, - "cuda_time_us": 104.512, - "pct_cuda_time": 1.1605272386109828, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.776, - "pct_cuda_time": 1.1523545115785112, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 49.486, - "cuda_time_us": 11.904, - "pct_cuda_time": 0.1321849763512816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.904, - "pct_cuda_time": 0.1321849763512816, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.84, - "cuda_time_us": 71.777, - "pct_cuda_time": 0.797029657893644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 71.777, - "pct_cuda_time": 0.797029657893644, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 985.482, - "cuda_time_us": 267.332, - "pct_cuda_time": 2.9685210095716403, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.264, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 712.448, - "cuda_time_us": 68.419, - "pct_cuda_time": 0.7597415908079918, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.071, - "cuda_time_us": 28.001, - "pct_cuda_time": 0.31093006744054397, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.001, - "pct_cuda_time": 0.31093006744054397, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 213.876, - "cuda_time_us": 4.705, - "pct_cuda_time": 0.05224549006491767, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.705, - "pct_cuda_time": 0.05224549006491767, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 306.235, - "cuda_time_us": 14.273, - "pct_cuda_time": 0.1584909414870499, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.913, - "pct_cuda_time": 0.03234667642063871, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.048, - "pct_cuda_time": 0.11157549079113553, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.014568774275275658, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.178, - "cuda_time_us": 21.44, - "pct_cuda_time": 0.23807509181548026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.44, - "pct_cuda_time": 0.23807509181548026, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.426, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 180.432, - "cuda_time_us": 189.761, - "pct_cuda_time": 2.1071533347946527, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.882, - "cuda_time_us": 104.76899999999999, - "pct_cuda_time": 1.1633810305231367, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.008183831281157133, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 104.032, - "pct_cuda_time": 1.1551971992419796, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 44.169, - "cuda_time_us": 12.256, - "pct_cuda_time": 0.13609367188855065, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.256, - "pct_cuda_time": 0.13609367188855065, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.839, - "cuda_time_us": 72.736, - "pct_cuda_time": 0.8076786323829651, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.736, - "pct_cuda_time": 0.8076786323829651, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 946.133, - "cuda_time_us": 267.999, - "pct_cuda_time": 2.975927543444818, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.023, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 664.894, - "cuda_time_us": 69.087, - "pct_cuda_time": 0.7671592289298548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 61.493, - "cuda_time_us": 28.223, - "pct_cuda_time": 0.31339521064870796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.223, - "pct_cuda_time": 0.31339521064870796, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 188.288, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 286.977, - "cuda_time_us": 14.528, - "pct_cuda_time": 0.1613225249018329, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.912, - "pct_cuda_time": 0.032335572171953285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.304, - "pct_cuda_time": 0.11441817845460395, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.014568774275275658, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.437, - "cuda_time_us": 21.696, - "pct_cuda_time": 0.2409177794789487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.696, - "pct_cuda_time": 0.2409177794789487, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.769, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 180.215, - "cuda_time_us": 189.888, - "pct_cuda_time": 2.1085635743777016, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.94, - "cuda_time_us": 105.024, - "pct_cuda_time": 1.1662126139379199, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 104.288, - "pct_cuda_time": 1.158039886905448, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.188, - "cuda_time_us": 11.84, - "pct_cuda_time": 0.13147430443541447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.84, - "pct_cuda_time": 0.13147430443541447, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.319, - "cuda_time_us": 73.024, - "pct_cuda_time": 0.8108766560043671, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 73.024, - "pct_cuda_time": 0.8108766560043671, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 950.376, - "cuda_time_us": 268.127, - "pct_cuda_time": 2.977348887276552, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.285, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 689.486, - "cuda_time_us": 69.184, - "pct_cuda_time": 0.7682363410523407, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 64.485, - "cuda_time_us": 27.936, - "pct_cuda_time": 0.3102082912759914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.936, - "pct_cuda_time": 0.3102082912759914, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 207.59, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 288.9, - "cuda_time_us": 14.944, - "pct_cuda_time": 0.1659418923549691, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.944, - "pct_cuda_time": 0.03269090812988684, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.336, - "pct_cuda_time": 0.1147735144125375, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.018477469812544736, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 73.6, - "cuda_time_us": 21.76, - "pct_cuda_time": 0.2416284513948158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.76, - "pct_cuda_time": 0.2416284513948158, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.841, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.407, - "cuda_time_us": 189.887, - "pct_cuda_time": 2.108552470129016, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.995, - "cuda_time_us": 105.31200000000001, - "pct_cuda_time": 1.1694106375593218, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.769, - "pct_cuda_time": 0.008539167239090687, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 104.543, - "pct_cuda_time": 1.1608714703202312, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.319, - "cuda_time_us": 11.711, - "pct_cuda_time": 0.13004185635499485, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.711, - "pct_cuda_time": 0.13004185635499485, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.231, - "cuda_time_us": 72.864, - "pct_cuda_time": 0.8090999762146994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.864, - "pct_cuda_time": 0.8090999762146994, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 887.753, - "cuda_time_us": 266.721, - "pct_cuda_time": 2.9617363136248462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.546, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 637.549, - "cuda_time_us": 68.737, - "pct_cuda_time": 0.7632727418899564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.511, - "cuda_time_us": 27.744, - "pct_cuda_time": 0.3080762755283901, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.744, - "pct_cuda_time": 0.3080762755283901, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 185.318, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 274.925, - "cuda_time_us": 14.529, - "pct_cuda_time": 0.16133362915051833, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.105, - "pct_cuda_time": 0.03447869216824003, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.92, - "pct_cuda_time": 0.11015414695940132, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01670079002287697, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.127, - "cuda_time_us": 21.824, - "pct_cuda_time": 0.24233912331068289, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.824, - "pct_cuda_time": 0.24233912331068289, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.66, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.037, - "cuda_time_us": 188.89600000000002, - "pct_cuda_time": 2.0975481596817613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.27, - "cuda_time_us": 104.0, - "pct_cuda_time": 1.154841863284046, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.008528062990405264, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.232, - "pct_cuda_time": 1.1463138002936408, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.666, - "cuda_time_us": 11.776, - "pct_cuda_time": 0.13076363251954737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.776, - "pct_cuda_time": 0.13076363251954737, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.447, - "cuda_time_us": 73.12, - "pct_cuda_time": 0.8119426638781677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 73.12, - "pct_cuda_time": 0.8119426638781677, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1020.261, - "cuda_time_us": 267.778, - "pct_cuda_time": 2.9734735044853395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.834, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 739.969, - "cuda_time_us": 68.38499999999999, - "pct_cuda_time": 0.7593640463526873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.932, - "cuda_time_us": 27.84, - "pct_cuda_time": 0.3091422834021908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.84, - "pct_cuda_time": 0.3091422834021908, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 271.082, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 275.426, - "cuda_time_us": 14.176999999999998, - "pct_cuda_time": 0.15742493361324922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.848, - "pct_cuda_time": 0.031624900256086184, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.049, - "pct_cuda_time": 0.11158659503982095, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.014213438317342106, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 73.95, - "cuda_time_us": 21.728, - "pct_cuda_time": 0.24127311543688224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.728, - "pct_cuda_time": 0.24127311543688224, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.379, - "cuda_time_us": 4.545, - "pct_cuda_time": 0.0504688102752499, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.545, - "pct_cuda_time": 0.0504688102752499, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.614, - "cuda_time_us": 190.208, - "pct_cuda_time": 2.1121169339570365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.766, - "cuda_time_us": 104.32000000000001, - "pct_cuda_time": 1.1583952228633816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.584, - "pct_cuda_time": 1.15022249583091, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.382, - "cuda_time_us": 11.936, - "pct_cuda_time": 0.13254031230921512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.936, - "pct_cuda_time": 0.13254031230921512, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.496, - "cuda_time_us": 73.952, - "pct_cuda_time": 0.82118139878444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 73.952, - "pct_cuda_time": 0.82118139878444, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 916.411, - "cuda_time_us": 269.664, - "pct_cuda_time": 2.9944161175060477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.106, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 662.235, - "cuda_time_us": 69.632, - "pct_cuda_time": 0.7732110444634106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.072, - "cuda_time_us": 28.864, - "pct_cuda_time": 0.32051303405606446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.864, - "pct_cuda_time": 0.32051303405606446, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 202.767, - "cuda_time_us": 4.896, - "pct_cuda_time": 0.05436640156383355, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.896, - "pct_cuda_time": 0.05436640156383355, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 286.422, - "cuda_time_us": 14.624, - "pct_cuda_time": 0.16238853277563356, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.944, - "pct_cuda_time": 0.03269090812988684, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.208, - "pct_cuda_time": 0.11335217058080328, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.01634545406494342, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.565, - "cuda_time_us": 21.248, - "pct_cuda_time": 0.23594307606787895, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.248, - "pct_cuda_time": 0.23594307606787895, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.337, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.242, - "cuda_time_us": 190.94400000000002, - "pct_cuda_time": 2.120289660989509, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.357, - "cuda_time_us": 105.72800000000001, - "pct_cuda_time": 1.1740300050124581, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 104.992, - "pct_cuda_time": 1.1658572779799863, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 41.202, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.13893635955201908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.13893635955201908, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.573, - "cuda_time_us": 72.704, - "pct_cuda_time": 0.8073232964250314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.704, - "pct_cuda_time": 0.8073232964250314, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 898.816, - "cuda_time_us": 268.097, - "pct_cuda_time": 2.977015759815989, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.393, - "cuda_time_us": 4.543, - "pct_cuda_time": 0.050446601777879053, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.543, - "pct_cuda_time": 0.050446601777879053, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 649.559, - "cuda_time_us": 69.506, - "pct_cuda_time": 0.7718119091290472, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.926, - "cuda_time_us": 28.0, - "pct_cuda_time": 0.3109189631918586, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.0, - "pct_cuda_time": 0.3109189631918586, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 195.368, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 280.427, - "cuda_time_us": 14.817999999999998, - "pct_cuda_time": 0.1645427570206057, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.944, - "pct_cuda_time": 0.03269090812988684, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.369, - "pct_cuda_time": 0.11513995461915646, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.016711894271562396, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.807, - "cuda_time_us": 22.112, - "pct_cuda_time": 0.24553714693208484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 22.112, - "pct_cuda_time": 0.24553714693208484, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.563, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 165.964, - "cuda_time_us": 189.6, - "pct_cuda_time": 2.105365550756299, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.072, - "cuda_time_us": 105.152, - "pct_cuda_time": 1.167633957769654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.008528062990405264, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 104.384, - "pct_cuda_time": 1.1591058947792487, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.082, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.12969762464574672, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.12969762464574672, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 50.626, - "cuda_time_us": 72.768, - "pct_cuda_time": 0.8080339683408986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.768, - "pct_cuda_time": 0.8080339683408986, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 958.902, - "cuda_time_us": 267.20099999999996, - "pct_cuda_time": 2.9670663529938497, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.531, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 669.762, - "cuda_time_us": 68.225, - "pct_cuda_time": 0.7575873665630195, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.445, - "cuda_time_us": 27.968, - "pct_cuda_time": 0.310563627233925, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.968, - "pct_cuda_time": 0.310563627233925, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 190.137, - "cuda_time_us": 4.545, - "pct_cuda_time": 0.0504688102752499, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.545, - "pct_cuda_time": 0.0504688102752499, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 293.812, - "cuda_time_us": 14.4, - "pct_cuda_time": 0.15990118107009868, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.848, - "pct_cuda_time": 0.031624900256086184, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.048, - "pct_cuda_time": 0.11157549079113553, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01670079002287697, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 74.9, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.23665374798374608, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.23665374798374608, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.618, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 206.001, - "cuda_time_us": 189.88799999999998, - "pct_cuda_time": 2.108563574377701, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.355, - "cuda_time_us": 104.64, - "pct_cuda_time": 1.1619485824427171, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.008528062990405264, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.872, - "pct_cuda_time": 1.153420519452312, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.101, - "cuda_time_us": 11.904, - "pct_cuda_time": 0.1321849763512816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.904, - "pct_cuda_time": 0.1321849763512816, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 83.362, - "cuda_time_us": 73.344, - "pct_cuda_time": 0.8144300155837025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 73.344, - "pct_cuda_time": 0.8144300155837025, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 898.546, - "cuda_time_us": 266.944, - "pct_cuda_time": 2.964212561081696, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.441, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 641.302, - "cuda_time_us": 68.83200000000001, - "pct_cuda_time": 0.7643276455150717, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.205, - "cuda_time_us": 28.256, - "pct_cuda_time": 0.313761650855327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.256, - "pct_cuda_time": 0.313761650855327, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.701, - "cuda_time_us": 4.704, - "pct_cuda_time": 0.052234385816232236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.704, - "pct_cuda_time": 0.052234385816232236, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 281.516, - "cuda_time_us": 14.176, - "pct_cuda_time": 0.1574138293645638, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.912, - "pct_cuda_time": 0.032335572171953285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.92, - "pct_cuda_time": 0.11015414695940132, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.014924110233209211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 67.051, - "cuda_time_us": 21.696, - "pct_cuda_time": 0.2409177794789487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.696, - "pct_cuda_time": 0.2409177794789487, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.936, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.127, - "cuda_time_us": 189.216, - "pct_cuda_time": 2.1011015192610967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 64.206, - "cuda_time_us": 104.16000000000001, - "pct_cuda_time": 1.156618543073714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.424, - "pct_cuda_time": 1.1484458160412423, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.301, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.13822568763615198, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.13822568763615198, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.202, - "cuda_time_us": 72.608, - "pct_cuda_time": 0.806257288551231, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.608, - "pct_cuda_time": 0.806257288551231, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 929.889, - "cuda_time_us": 264.704, - "pct_cuda_time": 2.9393390440263474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.357, - "cuda_time_us": 4.543, - "pct_cuda_time": 0.050446601777879053, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.543, - "pct_cuda_time": 0.050446601777879053, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 664.233, - "cuda_time_us": 68.16, - "pct_cuda_time": 0.756865590398467, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.615, - "cuda_time_us": 27.936, - "pct_cuda_time": 0.3102082912759914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.936, - "pct_cuda_time": 0.3102082912759914, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 180.354, - "cuda_time_us": 4.609, - "pct_cuda_time": 0.051179482191116996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.609, - "pct_cuda_time": 0.051179482191116996, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 300.825, - "cuda_time_us": 14.399, - "pct_cuda_time": 0.15989007682141324, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.911, - "pct_cuda_time": 0.03232446792326787, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.176, - "pct_cuda_time": 0.11299683462286973, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.014568774275275658, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 74.522, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.2355877401099454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.2355877401099454, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.77, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.115, - "cuda_time_us": 187.585, - "pct_cuda_time": 2.082990489655171, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.679, - "cuda_time_us": 103.68, - "pct_cuda_time": 1.1512885037047105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 102.944, - "pct_cuda_time": 1.1431157766722388, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.381, - "cuda_time_us": 11.776, - "pct_cuda_time": 0.13076363251954737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.776, - "pct_cuda_time": 0.13076363251954737, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.61, - "cuda_time_us": 72.129, - "pct_cuda_time": 0.8009383534309131, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.129, - "pct_cuda_time": 0.8009383534309131, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 933.709, - "cuda_time_us": 265.63, - "pct_cuda_time": 2.9496215783090496, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.478, - "cuda_time_us": 4.415, - "pct_cuda_time": 0.04902525794614484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.415, - "pct_cuda_time": 0.04902525794614484, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 678.885, - "cuda_time_us": 69.024, - "pct_cuda_time": 0.766459661262673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.1, - "cuda_time_us": 27.872, - "pct_cuda_time": 0.3094976193601243, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.872, - "pct_cuda_time": 0.3094976193601243, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 181.872, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 278.246, - "cuda_time_us": 14.592, - "pct_cuda_time": 0.1620331968177, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.88, - "pct_cuda_time": 0.031980236214019735, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.208, - "pct_cuda_time": 0.11335217058080328, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01670079002287697, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.732, - "cuda_time_us": 22.08, - "pct_cuda_time": 0.2451818109741513, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 22.08, - "pct_cuda_time": 0.2451818109741513, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.343, - "cuda_time_us": 4.383, - "pct_cuda_time": 0.04866992198821128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.383, - "pct_cuda_time": 0.04866992198821128, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 167.286, - "cuda_time_us": 187.808, - "pct_cuda_time": 2.0854667371120206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.256, - "cuda_time_us": 103.104, - "pct_cuda_time": 1.1448924564619065, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 102.368, - "pct_cuda_time": 1.1367197294294349, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.211, - "cuda_time_us": 12.416, - "pct_cuda_time": 0.13787035167821843, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.416, - "pct_cuda_time": 0.13787035167821843, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 50.645, - "cuda_time_us": 72.288, - "pct_cuda_time": 0.8027039289718954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.288, - "pct_cuda_time": 0.8027039289718954, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 978.899, - "cuda_time_us": 267.966, - "pct_cuda_time": 2.9755611032381992, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.51, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 726.725, - "cuda_time_us": 68.447, - "pct_cuda_time": 0.7600525097711837, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.146, - "cuda_time_us": 28.064, - "pct_cuda_time": 0.3116296351077256, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.064, - "pct_cuda_time": 0.3116296351077256, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 178.176, - "cuda_time_us": 4.671, - "pct_cuda_time": 0.05186794560961326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.671, - "pct_cuda_time": 0.05186794560961326, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 374.727, - "cuda_time_us": 14.304, - "pct_cuda_time": 0.15883517319629803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.944, - "pct_cuda_time": 0.03269090812988684, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.856, - "pct_cuda_time": 0.10944347504353422, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01670079002287697, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.702, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.23771975585754673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.23771975585754673, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.026, - "cuda_time_us": 4.383, - "pct_cuda_time": 0.04866992198821128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.383, - "pct_cuda_time": 0.04866992198821128, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.585, - "cuda_time_us": 190.59199999999998, - "pct_cuda_time": 2.1163809654522394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.234, - "cuda_time_us": 106.528, - "pct_cuda_time": 1.1829134039607967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 105.792, - "pct_cuda_time": 1.174740676928325, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.692, - "cuda_time_us": 11.52, - "pct_cuda_time": 0.12792094485607894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.52, - "pct_cuda_time": 0.12792094485607894, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 57.057, - "cuda_time_us": 72.544, - "pct_cuda_time": 0.8055466166353639, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.544, - "pct_cuda_time": 0.8055466166353639, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 918.788, - "cuda_time_us": 265.568, - "pct_cuda_time": 2.948933114890553, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.244, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 656.072, - "cuda_time_us": 68.352, - "pct_cuda_time": 0.7589976061460685, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.655, - "cuda_time_us": 28.032, - "pct_cuda_time": 0.3112742991497921, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.032, - "pct_cuda_time": 0.3112742991497921, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.74, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 290.646, - "cuda_time_us": 14.112, - "pct_cuda_time": 0.1567031574486967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.03482292387748816, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.696, - "pct_cuda_time": 0.10766679525386644, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.014213438317342106, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 75.287, - "cuda_time_us": 21.664, - "pct_cuda_time": 0.24056244352101516, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.664, - "pct_cuda_time": 0.24056244352101516, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.68, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.152, - "cuda_time_us": 188.192, - "pct_cuda_time": 2.089730768607223, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.862, - "cuda_time_us": 104.608, - "pct_cuda_time": 1.1615932464847836, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.872, - "pct_cuda_time": 1.153420519452312, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.807, - "cuda_time_us": 11.648, - "pct_cuda_time": 0.12934228868781314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.648, - "pct_cuda_time": 0.12934228868781314, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.386, - "cuda_time_us": 71.936, - "pct_cuda_time": 0.7987952334346263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 71.936, - "pct_cuda_time": 0.7987952334346263, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 902.48, - "cuda_time_us": 265.05600000000004, - "pct_cuda_time": 2.943247739563617, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.36, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 638.805, - "cuda_time_us": 68.096, - "pct_cuda_time": 0.7561549184826, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.652, - "cuda_time_us": 27.904, - "pct_cuda_time": 0.3098529553180579, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.904, - "pct_cuda_time": 0.3098529553180579, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 178.955, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 279.823, - "cuda_time_us": 14.368, - "pct_cuda_time": 0.15954584511216513, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.976, - "pct_cuda_time": 0.0330462440878204, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.08, - "pct_cuda_time": 0.11193082674906907, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.014568774275275658, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 74.673, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.23665374798374608, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.23665374798374608, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 40.68, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.341, - "cuda_time_us": 188.06400000000002, - "pct_cuda_time": 2.088309424775489, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.991, - "cuda_time_us": 103.68, - "pct_cuda_time": 1.1512885037047105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 102.944, - "pct_cuda_time": 1.1431157766722388, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.462, - "cuda_time_us": 12.128, - "pct_cuda_time": 0.13467232805681645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.128, - "pct_cuda_time": 0.13467232805681645, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.05, - "cuda_time_us": 72.256, - "pct_cuda_time": 0.8023485930139619, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.256, - "pct_cuda_time": 0.8023485930139619, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 917.188, - "cuda_time_us": 266.334, - "pct_cuda_time": 2.957438969383588, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.914, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 651.449, - "cuda_time_us": 68.31899999999999, - "pct_cuda_time": 0.7586311659394493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.409, - "cuda_time_us": 27.84, - "pct_cuda_time": 0.3091422834021908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.84, - "pct_cuda_time": 0.3091422834021908, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 192.759, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 279.988, - "cuda_time_us": 14.431, - "pct_cuda_time": 0.16024541277934679, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.911, - "pct_cuda_time": 0.03232446792326787, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.048, - "pct_cuda_time": 0.11157549079113553, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.01634545406494342, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.405, - "cuda_time_us": 21.536, - "pct_cuda_time": 0.23914109968928093, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.536, - "pct_cuda_time": 0.23914109968928093, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 38.124, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.458, - "cuda_time_us": 189.023, - "pct_cuda_time": 2.0989583992648098, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.282, - "cuda_time_us": 104.64, - "pct_cuda_time": 1.1619485824427171, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.904, - "pct_cuda_time": 1.1537758554102453, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.98, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.1289869527298796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.1289869527298796, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.796, - "cuda_time_us": 72.767, - "pct_cuda_time": 0.8080228640922131, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.767, - "pct_cuda_time": 0.8080228640922131, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 891.465, - "cuda_time_us": 265.79200000000003, - "pct_cuda_time": 2.9514204665960886, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.749, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 627.643, - "cuda_time_us": 68.735, - "pct_cuda_time": 0.7632505333925856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.423, - "cuda_time_us": 27.904, - "pct_cuda_time": 0.3098529553180579, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.904, - "pct_cuda_time": 0.3098529553180579, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 181.018, - "cuda_time_us": 4.992, - "pct_cuda_time": 0.05543240943763421, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.992, - "pct_cuda_time": 0.05543240943763421, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 282.291, - "cuda_time_us": 14.144, - "pct_cuda_time": 0.15705849340663028, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.848, - "pct_cuda_time": 0.031624900256086184, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.792, - "pct_cuda_time": 0.1087328031276671, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01670079002287697, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 64.025, - "cuda_time_us": 21.695, - "pct_cuda_time": 0.24090667523026327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.695, - "pct_cuda_time": 0.24090667523026327, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.294, - "cuda_time_us": 4.417, - "pct_cuda_time": 0.04904746644351568, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.417, - "pct_cuda_time": 0.04904746644351568, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 183.19, - "cuda_time_us": 188.16000000000003, - "pct_cuda_time": 2.0893754326492897, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.507, - "cuda_time_us": 103.808, - "pct_cuda_time": 1.1527098475364448, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.008528062990405264, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.04, - "pct_cuda_time": 1.1441817845460394, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.264, - "cuda_time_us": 11.808, - "pct_cuda_time": 0.13111896847748092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.808, - "pct_cuda_time": 0.13111896847748092, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.249, - "cuda_time_us": 72.544, - "pct_cuda_time": 0.8055466166353639, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.544, - "pct_cuda_time": 0.8055466166353639, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 899.835, - "cuda_time_us": 265.823, - "pct_cuda_time": 2.951764698305336, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.517, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 639.189, - "cuda_time_us": 68.96000000000001, - "pct_cuda_time": 0.765748989346806, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.505, - "cuda_time_us": 28.672, - "pct_cuda_time": 0.31838101830846316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.672, - "pct_cuda_time": 0.31838101830846316, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 188.508, - "cuda_time_us": 4.768, - "pct_cuda_time": 0.05294505773209934, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.768, - "pct_cuda_time": 0.05294505773209934, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 277.001, - "cuda_time_us": 13.92, - "pct_cuda_time": 0.1545711417010954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.816, - "pct_cuda_time": 0.03126956429815263, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.792, - "pct_cuda_time": 0.1087328031276671, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.014568774275275658, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 67.444, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.23985177160514803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.23985177160514803, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.21, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 177.884, - "cuda_time_us": 187.775, - "pct_cuda_time": 2.0851002969054013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.376, - "cuda_time_us": 103.168, - "pct_cuda_time": 1.1456031283777737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 102.432, - "pct_cuda_time": 1.137430401345302, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 40.225, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.1289869527298796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.1289869527298796, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 58.559, - "cuda_time_us": 72.991, - "pct_cuda_time": 0.8105102157977482, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.991, - "pct_cuda_time": 0.8105102157977482, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 893.012, - "cuda_time_us": 267.009, - "pct_cuda_time": 2.9649343372462487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.595, - "cuda_time_us": 4.639, - "pct_cuda_time": 0.05151260965167971, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.639, - "pct_cuda_time": 0.05151260965167971, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 629.609, - "cuda_time_us": 68.705, - "pct_cuda_time": 0.762917405932023, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.075, - "cuda_time_us": 28.001, - "pct_cuda_time": 0.31093006744054397, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.001, - "pct_cuda_time": 0.31093006744054397, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 188.642, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 275.808, - "cuda_time_us": 14.752999999999998, - "pct_cuda_time": 0.16382098085605318, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.036244267709222365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.145, - "pct_cuda_time": 0.11265260291362161, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.014924110233209211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.871, - "cuda_time_us": 21.311, - "pct_cuda_time": 0.23664264373506064, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.311, - "pct_cuda_time": 0.23664264373506064, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.899, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.591, - "cuda_time_us": 189.153, - "pct_cuda_time": 2.100401951593915, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 70.459, - "cuda_time_us": 105.02499999999999, - "pct_cuda_time": 1.166223718186605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.008183831281157133, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 104.288, - "pct_cuda_time": 1.158039886905448, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.395, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.1289869527298796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.1289869527298796, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.321, - "cuda_time_us": 72.512, - "pct_cuda_time": 0.8051912806774303, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.512, - "pct_cuda_time": 0.8051912806774303, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1035.71, - "cuda_time_us": 266.56100000000004, - "pct_cuda_time": 2.9599596338351795, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.826, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 772.86, - "cuda_time_us": 69.217, - "pct_cuda_time": 0.7686027812589596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.876, - "cuda_time_us": 28.416, - "pct_cuda_time": 0.31553833064499476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.416, - "pct_cuda_time": 0.31553833064499476, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 180.339, - "cuda_time_us": 4.736, - "pct_cuda_time": 0.05258972177416578, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.736, - "pct_cuda_time": 0.05258972177416578, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 400.701, - "cuda_time_us": 14.561, - "pct_cuda_time": 0.16168896510845185, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.912, - "pct_cuda_time": 0.032335572171953285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.144, - "pct_cuda_time": 0.11264149866493618, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.016711894271562396, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 77.516, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.23878576373134738, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.23878576373134738, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 36.221, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.04868102623689671, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.04868102623689671, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.417, - "cuda_time_us": 188.48000000000002, - "pct_cuda_time": 2.092928792228625, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.832, - "cuda_time_us": 104.352, - "pct_cuda_time": 1.1587505588213152, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.616, - "pct_cuda_time": 1.1505778317888433, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.3, - "cuda_time_us": 11.84, - "pct_cuda_time": 0.13147430443541447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.84, - "pct_cuda_time": 0.13147430443541447, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.975, - "cuda_time_us": 72.288, - "pct_cuda_time": 0.8027039289718954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.288, - "pct_cuda_time": 0.8027039289718954, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 899.23, - "cuda_time_us": 263.615, - "pct_cuda_time": 2.927246517207921, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.663, - "cuda_time_us": 4.32, - "pct_cuda_time": 0.047970354321029605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.32, - "pct_cuda_time": 0.047970354321029605, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 643.242, - "cuda_time_us": 67.681, - "pct_cuda_time": 0.7515466552781491, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.373, - "cuda_time_us": 27.488, - "pct_cuda_time": 0.3052335878649217, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.488, - "pct_cuda_time": 0.3052335878649217, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 190.46, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 275.449, - "cuda_time_us": 14.08, - "pct_cuda_time": 0.15634782149076315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.816, - "pct_cuda_time": 0.03126956429815263, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.76, - "pct_cuda_time": 0.10837746716973355, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01670079002287697, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.37, - "cuda_time_us": 21.633, - "pct_cuda_time": 0.24021821181176697, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.633, - "pct_cuda_time": 0.24021821181176697, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.29, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.411, - "cuda_time_us": 187.13400000000001, - "pct_cuda_time": 2.077982473498045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.546, - "cuda_time_us": 103.071, - "pct_cuda_time": 1.1445260162552877, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.008161622783786286, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 102.336, - "pct_cuda_time": 1.1363643934715013, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.516, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.1289869527298796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.1289869527298796, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.532, - "cuda_time_us": 72.447, - "pct_cuda_time": 0.8044695045128777, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.447, - "pct_cuda_time": 0.8044695045128777, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 942.366, - "cuda_time_us": 264.73600000000005, - "pct_cuda_time": 2.9396943799842816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.204, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 673.77, - "cuda_time_us": 68.513, - "pct_cuda_time": 0.7607853901844217, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 62.805, - "cuda_time_us": 28.0, - "pct_cuda_time": 0.3109189631918586, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.0, - "pct_cuda_time": 0.3109189631918586, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 181.255, - "cuda_time_us": 4.928, - "pct_cuda_time": 0.05472173752176711, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.928, - "pct_cuda_time": 0.05472173752176711, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 294.053, - "cuda_time_us": 14.177, - "pct_cuda_time": 0.15742493361324922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.041, - "pct_cuda_time": 0.033768020252372924, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.856, - "pct_cuda_time": 0.10944347504353422, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.014213438317342106, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 78.108, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.23771975585754673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.23771975585754673, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.996, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 182.117, - "cuda_time_us": 187.16700000000003, - "pct_cuda_time": 2.0783489137046645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 68.234, - "cuda_time_us": 103.072, - "pct_cuda_time": 1.1445371205039732, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 102.336, - "pct_cuda_time": 1.1363643934715013, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.108, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.12969762464574672, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.12969762464574672, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.107, - "cuda_time_us": 72.415, - "pct_cuda_time": 0.8041141685549442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.415, - "pct_cuda_time": 0.8041141685549442, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 919.754, - "cuda_time_us": 265.91900000000004, - "pct_cuda_time": 2.9528307061791375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.705, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 665.331, - "cuda_time_us": 67.711, - "pct_cuda_time": 0.7518797827387119, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 67.149, - "cuda_time_us": 27.679, - "pct_cuda_time": 0.3073544993638376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.679, - "pct_cuda_time": 0.3073544993638376, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 189.933, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 276.932, - "cuda_time_us": 14.272, - "pct_cuda_time": 0.15847983723836445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.944, - "pct_cuda_time": 0.03269090812988684, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.016, - "pct_cuda_time": 0.11122015483320197, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.014568774275275658, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 67.804, - "cuda_time_us": 21.12, - "pct_cuda_time": 0.23452173223614475, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.12, - "pct_cuda_time": 0.23452173223614475, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.076, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.859, - "cuda_time_us": 189.08800000000002, - "pct_cuda_time": 2.099680175429363, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.213, - "cuda_time_us": 104.64, - "pct_cuda_time": 1.1619485824427171, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.904, - "pct_cuda_time": 1.1537758554102453, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.202, - "cuda_time_us": 11.552, - "pct_cuda_time": 0.1282762808140125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.552, - "pct_cuda_time": 0.1282762808140125, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.197, - "cuda_time_us": 72.896, - "pct_cuda_time": 0.809455312172633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.896, - "pct_cuda_time": 0.809455312172633, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 908.013, - "cuda_time_us": 265.76, - "pct_cuda_time": 2.9510651306381543, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.561, - "cuda_time_us": 4.32, - "pct_cuda_time": 0.047970354321029605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.32, - "pct_cuda_time": 0.047970354321029605, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 645.003, - "cuda_time_us": 68.32, - "pct_cuda_time": 0.7586422701881348, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.339, - "cuda_time_us": 27.36, - "pct_cuda_time": 0.3038122440331875, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.36, - "pct_cuda_time": 0.3038122440331875, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 181.676, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 278.537, - "cuda_time_us": 14.784, - "pct_cuda_time": 0.1641652125653013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.03446758791955461, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.144, - "pct_cuda_time": 0.11264149866493618, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.01705612598081053, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.066, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.23985177160514803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.23985177160514803, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.308, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 174.148, - "cuda_time_us": 188.64, - "pct_cuda_time": 2.0947054720182927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.008, - "cuda_time_us": 104.768, - "pct_cuda_time": 1.1633699262744512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 104.032, - "pct_cuda_time": 1.1551971992419796, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.558, - "cuda_time_us": 11.807, - "pct_cuda_time": 0.1311078642287955, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.807, - "pct_cuda_time": 0.1311078642287955, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.87, - "cuda_time_us": 72.065, - "pct_cuda_time": 0.8002276815150459, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.065, - "pct_cuda_time": 0.8002276815150459, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 899.322, - "cuda_time_us": 266.014, - "pct_cuda_time": 2.953885609804252, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.235, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 640.935, - "cuda_time_us": 68.638, - "pct_cuda_time": 0.7621734212700996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.87, - "cuda_time_us": 27.84, - "pct_cuda_time": 0.3091422834021908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.84, - "pct_cuda_time": 0.3091422834021908, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 200.533, - "cuda_time_us": 4.992, - "pct_cuda_time": 0.05543240943763421, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.992, - "pct_cuda_time": 0.05543240943763421, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 268.54, - "cuda_time_us": 14.238, - "pct_cuda_time": 0.15810229278306007, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.008, - "pct_cuda_time": 0.03340158004575394, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.759, - "pct_cuda_time": 0.10836636292104813, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.471, - "pct_cuda_time": 0.016334349816258, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.66, - "cuda_time_us": 21.568, - "pct_cuda_time": 0.23949643564721446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.568, - "pct_cuda_time": 0.23949643564721446, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.693, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 167.512, - "cuda_time_us": 188.38400000000001, - "pct_cuda_time": 2.0918627843548245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.711, - "cuda_time_us": 103.936, - "pct_cuda_time": 1.154131191368179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.2, - "pct_cuda_time": 1.1459584643357072, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.544, - "cuda_time_us": 11.84, - "pct_cuda_time": 0.13147430443541447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.84, - "pct_cuda_time": 0.13147430443541447, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 50.178, - "cuda_time_us": 72.608, - "pct_cuda_time": 0.806257288551231, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.608, - "pct_cuda_time": 0.806257288551231, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 926.84, - "cuda_time_us": 265.373, - "pct_cuda_time": 2.9467677863968955, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.583, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 664.671, - "cuda_time_us": 67.519, - "pct_cuda_time": 0.7497477669911107, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.943, - "cuda_time_us": 27.296, - "pct_cuda_time": 0.30310157211732036, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.296, - "pct_cuda_time": 0.30310157211732036, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 190.134, - "cuda_time_us": 4.543, - "pct_cuda_time": 0.050446601777879053, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.543, - "pct_cuda_time": 0.050446601777879053, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 293.736, - "cuda_time_us": 14.112, - "pct_cuda_time": 0.1567031574486967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.88, - "pct_cuda_time": 0.031980236214019735, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.92, - "pct_cuda_time": 0.11015414695940132, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.014568774275275658, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.889, - "cuda_time_us": 21.568, - "pct_cuda_time": 0.23949643564721446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.568, - "pct_cuda_time": 0.23949643564721446, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.74, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.416, - "pct_cuda_time": 0.04903636219483027, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 177.134, - "cuda_time_us": 188.862, - "pct_cuda_time": 2.0971706152264566, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.256, - "cuda_time_us": 104.478, - "pct_cuda_time": 1.1601496941556784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.008161622783786286, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.743, - "pct_cuda_time": 1.151988071371892, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.857, - "cuda_time_us": 11.744, - "pct_cuda_time": 0.13040829656161382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.744, - "pct_cuda_time": 0.13040829656161382, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.688, - "cuda_time_us": 72.64, - "pct_cuda_time": 0.8066126245091645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.64, - "pct_cuda_time": 0.8066126245091645, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 994.545, - "cuda_time_us": 265.631, - "pct_cuda_time": 2.9496326825577346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.699, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 731.294, - "cuda_time_us": 68.54400000000001, - "pct_cuda_time": 0.7611296218936698, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.571, - "cuda_time_us": 28.16, - "pct_cuda_time": 0.3126956429815263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.16, - "pct_cuda_time": 0.3126956429815263, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.933, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 364.229, - "cuda_time_us": 14.432, - "pct_cuda_time": 0.16025651702803223, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.912, - "pct_cuda_time": 0.032335572171953285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.176, - "pct_cuda_time": 0.11299683462286973, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.014924110233209211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 70.057, - "cuda_time_us": 21.344, - "pct_cuda_time": 0.2370090839416796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.344, - "pct_cuda_time": 0.2370090839416796, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.915, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.04868102623689671, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.04868102623689671, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 174.762, - "cuda_time_us": 188.159, - "pct_cuda_time": 2.0893643284006043, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.867, - "cuda_time_us": 103.90400000000001, - "pct_cuda_time": 1.1537758554102455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.168, - "pct_cuda_time": 1.1456031283777737, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.427, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.13822568763615198, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.13822568763615198, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.852, - "cuda_time_us": 71.807, - "pct_cuda_time": 0.7973627853542067, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 71.807, - "pct_cuda_time": 0.7973627853542067, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 910.564, - "cuda_time_us": 265.986, - "pct_cuda_time": 2.95357469084106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.03, - "cuda_time_us": 4.481, - "pct_cuda_time": 0.04975813835938279, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.481, - "pct_cuda_time": 0.04975813835938279, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 649.136, - "cuda_time_us": 68.866, - "pct_cuda_time": 0.7647051899703761, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.196, - "cuda_time_us": 27.681, - "pct_cuda_time": 0.30737670786120846, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.681, - "pct_cuda_time": 0.30737670786120846, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 189.222, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.05116837794243158, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 284.266, - "cuda_time_us": 14.592, - "pct_cuda_time": 0.1620331968177, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.88, - "pct_cuda_time": 0.031980236214019735, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.208, - "pct_cuda_time": 0.11335217058080328, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01670079002287697, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 70.633, - "cuda_time_us": 21.985, - "pct_cuda_time": 0.24412690734903608, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.985, - "pct_cuda_time": 0.24412690734903608, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.403, - "cuda_time_us": 4.543, - "pct_cuda_time": 0.050446601777879053, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.543, - "pct_cuda_time": 0.050446601777879053, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 177.61, - "cuda_time_us": 188.096, - "pct_cuda_time": 2.088664760733422, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 67.642, - "cuda_time_us": 104.224, - "pct_cuda_time": 1.157329214989581, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.488, - "pct_cuda_time": 1.1491564879571092, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.373, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.12969762464574672, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.12969762464574672, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.492, - "cuda_time_us": 72.192, - "pct_cuda_time": 0.8016379210980947, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.192, - "pct_cuda_time": 0.8016379210980947, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 898.247, - "cuda_time_us": 265.312, - "pct_cuda_time": 2.9460904272270847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.703, - "cuda_time_us": 4.513, - "pct_cuda_time": 0.05011347431731634, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.513, - "pct_cuda_time": 0.05011347431731634, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 636.874, - "cuda_time_us": 68.607, - "pct_cuda_time": 0.7618291895608513, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.834, - "cuda_time_us": 27.936, - "pct_cuda_time": 0.3102082912759914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 27.936, - "pct_cuda_time": 0.3102082912759914, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 188.974, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.05152371390036512, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 278.828, - "cuda_time_us": 14.335, - "pct_cuda_time": 0.15917940490554616, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.912, - "pct_cuda_time": 0.032335572171953285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.856, - "pct_cuda_time": 0.10944347504353422, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.567, - "pct_cuda_time": 0.017400357690058654, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.015, - "cuda_time_us": 21.696, - "pct_cuda_time": 0.2409177794789487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.696, - "pct_cuda_time": 0.2409177794789487, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.806, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.04868102623689671, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.04868102623689671, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.668, - "cuda_time_us": 187.808, - "pct_cuda_time": 2.0854667371120206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 64.07, - "cuda_time_us": 104.03200000000001, - "pct_cuda_time": 1.1551971992419796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.296, - "pct_cuda_time": 1.147024472209508, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.781, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.12863161677194604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.12863161677194604, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.91, - "cuda_time_us": 72.192, - "pct_cuda_time": 0.8016379210980947, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.192, - "pct_cuda_time": 0.8016379210980947, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 916.262, - "cuda_time_us": 265.726, - "pct_cuda_time": 2.95068758618285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.114, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 670.126, - "cuda_time_us": 68.44800000000001, - "pct_cuda_time": 0.7600636140198691, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 49.733, - "cuda_time_us": 28.16, - "pct_cuda_time": 0.3126956429815263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.16, - "pct_cuda_time": 0.3126956429815263, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 179.713, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 317.543, - "cuda_time_us": 14.048, - "pct_cuda_time": 0.1559924855328296, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.816, - "pct_cuda_time": 0.03126956429815263, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 9.792, - "pct_cuda_time": 0.1087328031276671, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.015990118107009867, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 69.727, - "cuda_time_us": 21.76, - "pct_cuda_time": 0.2416284513948158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.76, - "pct_cuda_time": 0.2416284513948158, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.246, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.04939169815276382, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 167.49, - "cuda_time_us": 188.286, - "pct_cuda_time": 2.090774567983653, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.521, - "cuda_time_us": 104.224, - "pct_cuda_time": 1.157329214989581, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.488, - "pct_cuda_time": 1.1491564879571092, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.643, - "cuda_time_us": 11.647, - "pct_cuda_time": 0.12933118443912772, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.647, - "pct_cuda_time": 0.12933118443912772, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.549, - "cuda_time_us": 72.415, - "pct_cuda_time": 0.8041141685549442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.415, - "pct_cuda_time": 0.8041141685549442, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 901.925, - "cuda_time_us": 265.695, - "pct_cuda_time": 2.9503433544736017, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.95, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 649.866, - "cuda_time_us": 68.831, - "pct_cuda_time": 0.7643165412663863, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.799, - "cuda_time_us": 28.256, - "pct_cuda_time": 0.313761650855327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.256, - "pct_cuda_time": 0.313761650855327, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 189.752, - "cuda_time_us": 4.736, - "pct_cuda_time": 0.05258972177416578, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.736, - "pct_cuda_time": 0.05258972177416578, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 285.664, - "cuda_time_us": 14.431000000000001, - "pct_cuda_time": 0.1602454127793468, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.912, - "pct_cuda_time": 0.032335572171953285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.112, - "pct_cuda_time": 0.11228616270700263, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.407, - "pct_cuda_time": 0.015623677900390893, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.877, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.23771975585754673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.23771975585754673, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.402, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.04868102623689671, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.04868102623689671, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.979, - "cuda_time_us": 187.936, - "pct_cuda_time": 2.086888080943755, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.417, - "cuda_time_us": 103.52000000000001, - "pct_cuda_time": 1.1495118239150428, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 102.784, - "pct_cuda_time": 1.1413390968825712, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.623, - "cuda_time_us": 11.488, - "pct_cuda_time": 0.1275656088981454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.488, - "pct_cuda_time": 0.1275656088981454, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.573, - "cuda_time_us": 72.928, - "pct_cuda_time": 0.8098106481305665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.928, - "pct_cuda_time": 0.8098106481305665, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 873.553, - "cuda_time_us": 266.752, - "pct_cuda_time": 2.9620805453340946, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.627, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.512, - "pct_cuda_time": 0.05010237006863092, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 622.747, - "cuda_time_us": 69.47200000000001, - "pct_cuda_time": 0.7714343646737428, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 49.792, - "cuda_time_us": 28.608, - "pct_cuda_time": 0.31767034639259606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 28.608, - "pct_cuda_time": 0.31767034639259606, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[256, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 174.071, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.050457706026564464, - "trace": "_C::rotary_embedding(int64[256], bfloat16[256, 4096], bfloat16[256, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 275.378, - "cuda_time_us": 14.496, - "pct_cuda_time": 0.16096718894389933, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.944, - "pct_cuda_time": 0.03269090812988684, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[256], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.048, - "pct_cuda_time": 0.11157549079113553, - "trace": "_vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01670079002287697, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], None, None, bfloat16[256, 32, 128], int32[3], int32[3], None, None, None, 128, 128, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[256, 32, 128], bfloat16[256, 8, 128], bfloat16[256, 8, 128], bfloat16[256, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.005, - "cuda_time_us": 21.824, - "pct_cuda_time": 0.24233912331068289, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.824, - "pct_cuda_time": 0.24233912331068289, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[256, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.504, - "cuda_time_us": 4.352, - "pct_cuda_time": 0.04832569027896316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.352, - "pct_cuda_time": 0.04832569027896316, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.271, - "cuda_time_us": 188.416, - "pct_cuda_time": 2.092218120312758, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.035, - "cuda_time_us": 104.48, - "pct_cuda_time": 1.1601719026530493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.008528062990405264, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 103.712, - "pct_cuda_time": 1.151643839662644, - "trace": "mm(bfloat16[256, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[256, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[256, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.54, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.12863161677194604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.12863161677194604, - "trace": "_C::silu_and_mul(bfloat16[256, 14336], bfloat16[256, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.361, - "cuda_time_us": 72.352, - "pct_cuda_time": 0.8034146008877625, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 72.352, - "pct_cuda_time": 0.8034146008877625, - "trace": "mm(bfloat16[256, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[256, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[256, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.985, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.48, - "pct_cuda_time": 0.04974703411069737, - "trace": "_C::fused_add_rms_norm(bfloat16[256, 4096], bfloat16[256, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 163.55, - "cuda_time_us": 349.632, - "pct_cuda_time": 3.882400676381996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.03482292387748816, - "trace": "index_select(bfloat16[256, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.00817272703247171, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 345.76, - "pct_cuda_time": 3.839405025472036, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 1209.6, - "cuda_time_us": 117.05600000000001, - "pct_cuda_time": 1.2998189341209356, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.527, - "pct_cuda_time": 0.028060436428065237, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.177, - "pct_cuda_time": 0.024173949388167002, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.024518181097415135, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.008528062990405264, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.767, - "pct_cuda_time": 0.00851695874171984, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.801, - "pct_cuda_time": 0.008894503197024239, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.8, - "pct_cuda_time": 0.008883398948338816, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.192, - "pct_cuda_time": 0.0465490104892954, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.576, - "pct_cuda_time": 0.05081304198449803, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.728, - "pct_cuda_time": 0.3745240996619645, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.808, - "pct_cuda_time": 0.3087869474442573, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 1.92, - "pct_cuda_time": 0.02132015747601316, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.736, - "pct_cuda_time": 0.05258972177416578, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 27.456, - "pct_cuda_time": 0.30487825190698814, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.028782212592617765, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - }, - "decode_1": { - "metadata": { - "num_running_seqs": 2 - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 6306.9850000000015, - "pct_cuda_time": 93.12878618070069, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 3.552, - "pct_cuda_time": 0.05244874508403758, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.552, - "pct_cuda_time": 0.05244874508403758, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 6300.298, - "pct_cuda_time": 93.03004610232877, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 206.11199999999994, - "pct_cuda_time": 3.043444748525099, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.544, - "pct_cuda_time": 0.06709659281021024, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 201.56799999999996, - "pct_cuda_time": 2.9763481557148888, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 1829.9759999999999, - "pct_cuda_time": 27.021380837248522, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 667.9950000000001, - "pct_cuda_time": 9.863597824440228, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 667.9950000000001, - "pct_cuda_time": 9.863597824440228, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 120.99200000000002, - "pct_cuda_time": 1.786564911376091, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 120.99200000000002, - "pct_cuda_time": 1.786564911376091, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 478.56600000000014, - "pct_cuda_time": 7.06649384568906, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 79.45700000000002, - "pct_cuda_time": 1.1732601177202633, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 351.20200000000006, - "pct_cuda_time": 5.185840138233156, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 47.907, - "pct_cuda_time": 0.7073935897356386, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 562.4229999999999, - "pct_cuda_time": 8.304724255743148, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cuda_time_us": 493.65999999999997, - "pct_cuda_time": 7.289371480345156, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cuda_time_us": 68.763, - "pct_cuda_time": 1.0153527753979945, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 4264.210000000001, - "pct_cuda_time": 62.96522051655516, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 2590.9700000000003, - "pct_cuda_time": 38.25819962004191, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2590.9700000000003, - "pct_cuda_time": 38.25819962004191, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 285.914, - "pct_cuda_time": 4.221799127803356, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 285.914, - "pct_cuda_time": 4.221799127803356, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 1387.326, - "pct_cuda_time": 20.485221768709888, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1387.326, - "pct_cuda_time": 20.485221768709888, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04629133328785411, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04629133328785411, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 345.951, - "pct_cuda_time": 5.10830400072294, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.008, - "pct_cuda_time": 0.04441605439549129, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010867757990386166, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 342.207, - "pct_cuda_time": 5.053020188337063, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 119.38999999999999, - "pct_cuda_time": 1.7629098185763643, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 11.198, - "pct_cuda_time": 0.16534939398959822, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.128, - "pct_cuda_time": 0.06095394698955719, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.608, - "pct_cuda_time": 0.06804161524415686, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 34.304, - "pct_cuda_time": 0.50653202459539, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.904, - "pct_cuda_time": 0.4120297812007277, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 1.856, - "pct_cuda_time": 0.027405650584452074, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.736, - "pct_cuda_time": 0.06993166011205011, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 28.16, - "pct_cuda_time": 0.4158098709365142, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.496, - "pct_cuda_time": 0.0368558749239183, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 30573.472, - "cuda_time_us": 6306.9850000000015, - "pct_cuda_time": 93.12878618070069, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 102.367, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.05244874508403758, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.05244874508403758, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[2]) <- embedding(bfloat16[128256, 4096], int64[2], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1481.538, - "cuda_time_us": 200.899, - "pct_cuda_time": 2.9664697180850412, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 94.214, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.06709659281021024, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.544, - "pct_cuda_time": 0.06709659281021024, - "trace": "_C::rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 1057.757, - "cuda_time_us": 59.97, - "pct_cuda_time": 0.8855155525590467, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 154.632, - "cuda_time_us": 23.233, - "pct_cuda_time": 0.3430579094981546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 23.233, - "pct_cuda_time": 0.3430579094981546, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 344.518, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05481130116890415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05481130116890415, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 365.232, - "cuda_time_us": 14.848999999999998, - "pct_cuda_time": 0.21925997065114694, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.433, - "pct_cuda_time": 0.03592561846550209, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.072, - "pct_cuda_time": 0.1634888810727658, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019845471112879088, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 100.739, - "cuda_time_us": 18.176000000000002, - "pct_cuda_time": 0.26838637124084097, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.032, - "pct_cuda_time": 0.23672811970362911, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 52.392, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048668655348251086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048668655348251086, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 233.353, - "cuda_time_us": 133.089, - "pct_cuda_time": 1.965188917367533, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 89.673, - "cuda_time_us": 80.993, - "pct_cuda_time": 1.195940656134982, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.993, - "pct_cuda_time": 1.195940656134982, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 50.898, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12899556223371406, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12899556223371406, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 63.478, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.6402526989988371, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.6402526989988371, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 982.45, - "cuda_time_us": 197.214, - "pct_cuda_time": 2.9120570982554583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.747, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04819614413127778, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04819614413127778, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 697.627, - "cuda_time_us": 57.12, - "pct_cuda_time": 0.8434325222973611, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.913, - "cuda_time_us": 20.96, - "pct_cuda_time": 0.30949484711751907, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.96, - "pct_cuda_time": 0.30949484711751907, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 210.48, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05481130116890415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05481130116890415, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 289.706, - "cuda_time_us": 14.943999999999999, - "pct_cuda_time": 0.2206627383265365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.035910852489971674, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.008, - "pct_cuda_time": 0.16254385863881915, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 81.259, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.2584636356844014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.22680538414718956, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.428, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 188.595, - "cuda_time_us": 133.63, - "pct_cuda_time": 1.973177310129488, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 68.128, - "cuda_time_us": 80.607, - "pct_cuda_time": 1.1902409895802415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.607, - "pct_cuda_time": 1.1902409895802415, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 40.533, - "cuda_time_us": 9.407, - "pct_cuda_time": 0.13890353181462317, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.407, - "pct_cuda_time": 0.13890353181462317, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 57.829, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.6440327887346237, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.6440327887346237, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 923.056, - "cuda_time_us": 197.248, - "pct_cuda_time": 2.9125591414234924, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.786, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 649.077, - "cuda_time_us": 57.376, - "pct_cuda_time": 0.8472126120331476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.527, - "cuda_time_us": 20.767, - "pct_cuda_time": 0.3066450138401488, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.767, - "pct_cuda_time": 0.3066450138401488, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 201.932, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05670134603679739, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05670134603679739, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 274.857, - "cuda_time_us": 15.265, - "pct_cuda_time": 0.22540261647180002, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.037800897357864925, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.201, - "pct_cuda_time": 0.16539369191618947, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.771, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.2584636356844014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.264, - "pct_cuda_time": 0.2253878504962696, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03307578518813181, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 36.99, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.748, - "cuda_time_us": 133.536, - "pct_cuda_time": 1.9717893084296292, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.01, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1954533789424782, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1954533789424782, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.619, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1294680734506874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1294680734506874, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.476, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6468678560364636, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6468678560364636, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 882.084, - "cuda_time_us": 197.825, - "pct_cuda_time": 2.921079109304542, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.914, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 620.417, - "cuda_time_us": 57.504000000000005, - "pct_cuda_time": 0.8491026569010409, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.624, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.3142199592872522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.3142199592872522, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 185.851, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.05575632360285076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.05575632360285076, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 264.601, - "cuda_time_us": 14.848, - "pct_cuda_time": 0.2192452046756166, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.976, - "pct_cuda_time": 0.16207134742184587, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021735515980772332, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.751, - "cuda_time_us": 17.6, - "pct_cuda_time": 0.25988116933532135, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.456, - "pct_cuda_time": 0.2282229177981095, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.432, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.050100954974701434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.050100954974701434, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 180.177, - "cuda_time_us": 133.792, - "pct_cuda_time": 1.9755693981654154, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.527, - "cuda_time_us": 81.088, - "pct_cuda_time": 1.1973434238103713, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.088, - "pct_cuda_time": 1.1973434238103713, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 46.102, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13277565196950056, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13277565196950056, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.358, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6454503223855437, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6454503223855437, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 912.812, - "cuda_time_us": 197.98200000000003, - "pct_cuda_time": 2.9233973674628184, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.367, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 648.089, - "cuda_time_us": 57.472, - "pct_cuda_time": 0.8486301456840676, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 59.636, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.31611000415514545, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.31611000415514545, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 195.624, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.053866278734957515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.053866278734957515, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 273.201, - "cuda_time_us": 14.848, - "pct_cuda_time": 0.2192452046756166, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.035910852489971674, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.944, - "pct_cuda_time": 0.16159883620487256, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021735515980772332, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.35, - "cuda_time_us": 17.567999999999998, - "pct_cuda_time": 0.259408658118348, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.22775040658113616, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.945, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.563, - "cuda_time_us": 134.11, - "pct_cuda_time": 1.980264978384088, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.722, - "cuda_time_us": 81.184, - "pct_cuda_time": 1.1987609574612914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.184, - "pct_cuda_time": 1.1987609574612914, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.069, - "cuda_time_us": 8.959, - "pct_cuda_time": 0.13228837477699681, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.959, - "pct_cuda_time": 0.13228837477699681, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 61.938, - "cuda_time_us": 43.967, - "pct_cuda_time": 0.6492156461457996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.967, - "pct_cuda_time": 0.6492156461457996, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 967.167, - "cuda_time_us": 197.21299999999997, - "pct_cuda_time": 2.912042332279927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.52, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 677.618, - "cuda_time_us": 57.373999999999995, - "pct_cuda_time": 0.8471830800820868, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.046, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.31091238076843897, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.31091238076843897, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.079, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.05575632360285076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.05575632360285076, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 319.577, - "cuda_time_us": 14.942, - "pct_cuda_time": 0.22063320637547568, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.463, - "pct_cuda_time": 0.03636859773141457, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.008, - "pct_cuda_time": 0.16254385863881915, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.471, - "pct_cuda_time": 0.02172075000524192, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.482, - "cuda_time_us": 17.599999999999998, - "pct_cuda_time": 0.25988116933532135, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.296, - "pct_cuda_time": 0.2258603617132429, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03402080762207843, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.587, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.459, - "cuda_time_us": 133.503, - "pct_cuda_time": 1.9713020312371252, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.029, - "cuda_time_us": 80.832, - "pct_cuda_time": 1.193563334074585, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.832, - "pct_cuda_time": 1.193563334074585, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.938, - "cuda_time_us": 8.991, - "pct_cuda_time": 0.13276088599397012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.991, - "pct_cuda_time": 0.13276088599397012, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.215, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6449778111685703, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6449778111685703, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 876.717, - "cuda_time_us": 198.24, - "pct_cuda_time": 2.9272069891496653, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.057, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 620.245, - "cuda_time_us": 57.568, - "pct_cuda_time": 0.8500476793349875, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.058, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.3142199592872522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.3142199592872522, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 182.204, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.056228834819824075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.056228834819824075, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 265.414, - "cuda_time_us": 15.168, - "pct_cuda_time": 0.22397031684534968, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.624, - "pct_cuda_time": 0.03874591979181155, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.072, - "pct_cuda_time": 0.1634888810727658, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021735515980772332, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.599, - "cuda_time_us": 17.312, - "pct_cuda_time": 0.25562856838256154, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.169, - "pct_cuda_time": 0.2239850828208801, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.03164348556168145, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.479, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.541, - "cuda_time_us": 134.36800000000002, - "pct_cuda_time": 1.9840746000709355, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.635, - "cuda_time_us": 82.016, - "pct_cuda_time": 1.2110462491025975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.016, - "pct_cuda_time": 1.2110462491025975, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.122, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.704, - "cuda_time_us": 43.52, - "pct_cuda_time": 0.6426152550837038, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.52, - "pct_cuda_time": 0.6426152550837038, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 896.685, - "cuda_time_us": 197.31, - "pct_cuda_time": 2.9134746319063782, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.81, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045833588046411224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045833588046411224, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 643.417, - "cuda_time_us": 57.792, - "pct_cuda_time": 0.8533552578538007, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 59.891, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.3142199592872522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.28, - "pct_cuda_time": 0.3142199592872522, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 200.986, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.058118879687717326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.058118879687717326, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 267.327, - "cuda_time_us": 15.040000000000001, - "pct_cuda_time": 0.22208027197745647, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.037800897357864925, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.976, - "pct_cuda_time": 0.16207134742184587, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.541, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2589361469013748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.392, - "pct_cuda_time": 0.22727789536416285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.032, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.73, - "cuda_time_us": 133.246, - "pct_cuda_time": 1.9675071755258087, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.13, - "cuda_time_us": 80.607, - "pct_cuda_time": 1.1902409895802415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.607, - "pct_cuda_time": 1.1902409895802415, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.466, - "cuda_time_us": 8.831, - "pct_cuda_time": 0.13039832990910358, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.831, - "pct_cuda_time": 0.13039832990910358, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 57.234, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6468678560364636, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6468678560364636, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 896.432, - "cuda_time_us": 198.495, - "pct_cuda_time": 2.930972312909921, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.363, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045833588046411224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045833588046411224, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 632.449, - "cuda_time_us": 57.535000000000004, - "pct_cuda_time": 0.8495604021424839, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.423, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.3175275378060654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.3175275378060654, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 192.903, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05433878995193083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05433878995193083, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 270.379, - "cuda_time_us": 14.943, - "pct_cuda_time": 0.22064797235100608, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.591, - "pct_cuda_time": 0.03825864259930782, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.848, - "pct_cuda_time": 0.16018130255395263, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.865, - "cuda_time_us": 17.408, - "pct_cuda_time": 0.2570461020334815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.296, - "pct_cuda_time": 0.2258603617132429, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.03118574032023856, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.357, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048668655348251086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048668655348251086, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 177.763, - "cuda_time_us": 134.56, - "pct_cuda_time": 1.986909667372775, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.286, - "cuda_time_us": 81.312, - "pct_cuda_time": 1.2006510023291845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.312, - "pct_cuda_time": 1.2006510023291845, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.316, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.13183062953555394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.13183062953555394, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.097, - "cuda_time_us": 44.32, - "pct_cuda_time": 0.6544280355080365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.32, - "pct_cuda_time": 0.6544280355080365, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 934.111, - "cuda_time_us": 195.61400000000003, - "pct_cuda_time": 2.888431537406793, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.311, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04629133328785411, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04629133328785411, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 680.042, - "cuda_time_us": 56.703, - "pct_cuda_time": 0.8372751105011776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 86.623, - "cuda_time_us": 20.48, - "pct_cuda_time": 0.3024071788629194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.48, - "pct_cuda_time": 0.3024071788629194, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 210.205, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.05764636847074401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.05764636847074401, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 265.388, - "cuda_time_us": 14.975999999999999, - "pct_cuda_time": 0.2211352495435098, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.071, - "pct_cuda_time": 0.1634741150972354, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.022222793173276058, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 62.879, - "cuda_time_us": 17.343, - "pct_cuda_time": 0.25608631362400447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.199, - "pct_cuda_time": 0.22442806208679256, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.795, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.459, - "cuda_time_us": 132.57600000000002, - "pct_cuda_time": 1.95761397192043, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.037, - "cuda_time_us": 80.736, - "pct_cuda_time": 1.1921458004236651, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.736, - "pct_cuda_time": 1.1921458004236651, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.442, - "cuda_time_us": 9.12, - "pct_cuda_time": 0.13466569683739377, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.12, - "pct_cuda_time": 0.13466569683739377, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.942, - "cuda_time_us": 42.72, - "pct_cuda_time": 0.6308024746593709, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.72, - "pct_cuda_time": 0.6308024746593709, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 891.038, - "cuda_time_us": 196.03300000000002, - "pct_cuda_time": 2.8946184811540374, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.491, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 628.471, - "cuda_time_us": 56.545, - "pct_cuda_time": 0.8349420863673719, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.069, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.3052422461647593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.3052422461647593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 189.872, - "cuda_time_us": 3.681, - "pct_cuda_time": 0.054353555927461245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.681, - "pct_cuda_time": 0.054353555927461245, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 263.186, - "cuda_time_us": 14.912, - "pct_cuda_time": 0.2201902271095632, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.035910852489971674, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.976, - "pct_cuda_time": 0.16207134742184587, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.253, - "cuda_time_us": 17.28, - "pct_cuda_time": 0.25515605716558826, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.168, - "pct_cuda_time": 0.22397031684534968, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.03118574032023856, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.825, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04819614413127778, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04819614413127778, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.237, - "cuda_time_us": 133.056, - "pct_cuda_time": 1.9647016401750297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.804, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1940358452915583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1940358452915583, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.471, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12994058466766067, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12994058466766067, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.833, - "cuda_time_us": 43.392, - "pct_cuda_time": 0.6407252102158105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.392, - "pct_cuda_time": 0.6407252102158105, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 938.786, - "cuda_time_us": 196.959, - "pct_cuda_time": 2.9082917744952024, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 36.075, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 676.937, - "cuda_time_us": 56.607, - "pct_cuda_time": 0.8358575768502577, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.612, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.304769734947786, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.304769734947786, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 201.425, - "cuda_time_us": 3.711, - "pct_cuda_time": 0.05479653519337372, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.711, - "pct_cuda_time": 0.05479653519337372, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 271.565, - "cuda_time_us": 15.008000000000001, - "pct_cuda_time": 0.22160776076048314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.624, - "pct_cuda_time": 0.03874591979181155, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.88, - "pct_cuda_time": 0.16065381377092594, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.622, - "cuda_time_us": 17.247999999999998, - "pct_cuda_time": 0.2546835459486149, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.104, - "pct_cuda_time": 0.22302529441140304, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.389, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048668655348251086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048668655348251086, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.734, - "cuda_time_us": 133.856, - "pct_cuda_time": 1.9765144205993619, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.24, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1940358452915583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1940358452915583, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.417, - "cuda_time_us": 8.833, - "pct_cuda_time": 0.13042786186016442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.833, - "pct_cuda_time": 0.13042786186016442, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.697, - "cuda_time_us": 44.159, - "pct_cuda_time": 0.6520507134476395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.159, - "pct_cuda_time": 0.6520507134476395, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 957.677, - "cuda_time_us": 195.616, - "pct_cuda_time": 2.8884610693578536, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.439, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 698.064, - "cuda_time_us": 56.896, - "pct_cuda_time": 0.840124943778548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.527, - "cuda_time_us": 20.735, - "pct_cuda_time": 0.30617250262317547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.735, - "pct_cuda_time": 0.30617250262317547, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.469, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.053866278734957515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.053866278734957515, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 344.781, - "cuda_time_us": 14.977, - "pct_cuda_time": 0.2211500155190402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.0368558749239183, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.976, - "pct_cuda_time": 0.16207134742184587, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.022222793173276058, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 64.091, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2589361469013748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.392, - "pct_cuda_time": 0.22727789536416285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.338, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.256, - "cuda_time_us": 132.352, - "pct_cuda_time": 1.9543063934016165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.603, - "cuda_time_us": 81.024, - "pct_cuda_time": 1.1963984013764248, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.024, - "pct_cuda_time": 1.1963984013764248, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.628, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.12757802858279413, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.12757802858279413, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.652, - "cuda_time_us": 42.688, - "pct_cuda_time": 0.6303299634423976, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.688, - "pct_cuda_time": 0.6303299634423976, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 884.519, - "cuda_time_us": 196.445, - "pct_cuda_time": 2.900702063072568, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.257, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 625.621, - "cuda_time_us": 56.67, - "pct_cuda_time": 0.8367878333086739, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.555, - "cuda_time_us": 20.415, - "pct_cuda_time": 0.3014473904534423, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.415, - "pct_cuda_time": 0.3014473904534423, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 188.203, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.05764636847074401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.05764636847074401, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 271.205, - "cuda_time_us": 14.816, - "pct_cuda_time": 0.21877269345864325, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.912, - "pct_cuda_time": 0.16112632498789925, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.546, - "cuda_time_us": 17.535, - "pct_cuda_time": 0.2589213809258443, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.392, - "pct_cuda_time": 0.22727789536416285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.03164348556168145, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.649, - "cuda_time_us": 3.231, - "pct_cuda_time": 0.04770886693877405, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.231, - "pct_cuda_time": 0.04770886693877405, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.278, - "cuda_time_us": 133.344, - "pct_cuda_time": 1.9689542411277892, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.744, - "cuda_time_us": 81.472, - "pct_cuda_time": 1.203013558414051, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.472, - "pct_cuda_time": 1.203013558414051, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.95, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.844, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.635527586829104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.635527586829104, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 916.413, - "cuda_time_us": 198.20600000000002, - "pct_cuda_time": 2.9267049459816312, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.065, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 655.157, - "cuda_time_us": 57.311, - "pct_cuda_time": 0.8462528236236706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 65.116, - "cuda_time_us": 21.184, - "pct_cuda_time": 0.31280242563633226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.184, - "pct_cuda_time": 0.31280242563633226, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 196.183, - "cuda_time_us": 3.679, - "pct_cuda_time": 0.054324023976400404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.679, - "pct_cuda_time": 0.054324023976400404, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 277.562, - "cuda_time_us": 14.943999999999999, - "pct_cuda_time": 0.2206627383265365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.035910852489971674, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.008, - "pct_cuda_time": 0.16254385863881915, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.43, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.2584636356844014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.22680538414718956, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.044, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.482, - "cuda_time_us": 134.559, - "pct_cuda_time": 1.9868949013972446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.727, - "cuda_time_us": 81.823, - "pct_cuda_time": 1.2081964158252272, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.823, - "pct_cuda_time": 1.2081964158252272, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.61, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.13183062953555394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.13183062953555394, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.694, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6468678560364636, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6468678560364636, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 934.183, - "cuda_time_us": 195.745, - "pct_cuda_time": 2.8903658802012773, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.766, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.049613677782197704, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.049613677782197704, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 658.721, - "cuda_time_us": 56.674, - "pct_cuda_time": 0.8368468972107955, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.524, - "cuda_time_us": 20.352, - "pct_cuda_time": 0.30051713399502616, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.352, - "pct_cuda_time": 0.30051713399502616, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 193.414, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05481130116890415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05481130116890415, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 271.576, - "cuda_time_us": 15.042, - "pct_cuda_time": 0.22210980392851726, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.593, - "pct_cuda_time": 0.038288174550368655, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.977, - "pct_cuda_time": 0.16208611339737627, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021735515980772332, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 86.939, - "cuda_time_us": 17.567999999999998, - "pct_cuda_time": 0.259408658118348, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.22775040658113616, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.266, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.049613677782197704, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.049613677782197704, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 186.825, - "cuda_time_us": 132.351, - "pct_cuda_time": 1.9542916274260862, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 69.054, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1893107331218251, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1893107331218251, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.672, - "cuda_time_us": 8.639, - "pct_cuda_time": 0.1275632626072637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.639, - "pct_cuda_time": 0.1275632626072637, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.912, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6374176316969973, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6374176316969973, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 905.764, - "cuda_time_us": 196.54200000000003, - "pct_cuda_time": 2.902134362699019, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.706, - "cuda_time_us": 3.103, - "pct_cuda_time": 0.0458188220708808, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.103, - "pct_cuda_time": 0.0458188220708808, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 638.734, - "cuda_time_us": 56.607, - "pct_cuda_time": 0.8358575768502577, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.44, - "cuda_time_us": 20.447, - "pct_cuda_time": 0.30191990167041566, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.447, - "pct_cuda_time": 0.30191990167041566, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 187.191, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.058118879687717326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.058118879687717326, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 270.568, - "cuda_time_us": 14.913, - "pct_cuda_time": 0.22020499308509361, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.0368558749239183, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.912, - "pct_cuda_time": 0.16112632498789925, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.022222793173276058, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.212, - "cuda_time_us": 17.311, - "pct_cuda_time": 0.25561380240703113, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.168, - "pct_cuda_time": 0.22397031684534968, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.03164348556168145, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 35.823, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.979, - "cuda_time_us": 133.66400000000002, - "pct_cuda_time": 1.9736793532975225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.183, - "cuda_time_us": 82.016, - "pct_cuda_time": 1.2110462491025975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.016, - "pct_cuda_time": 1.2110462491025975, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.146, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.518, - "cuda_time_us": 42.816, - "pct_cuda_time": 0.6322200083102909, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.816, - "pct_cuda_time": 0.6322200083102909, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 862.168, - "cuda_time_us": 195.93599999999998, - "pct_cuda_time": 2.8931861815275863, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.021, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04819614413127778, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04819614413127778, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 602.08, - "cuda_time_us": 56.705, - "pct_cuda_time": 0.8373046424522385, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.395, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30429722373081264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30429722373081264, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.508, - "cuda_time_us": 3.713, - "pct_cuda_time": 0.05482606714443456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.713, - "pct_cuda_time": 0.05482606714443456, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 256.893, - "cuda_time_us": 14.88, - "pct_cuda_time": 0.21971771589258987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.035910852489971674, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.944, - "pct_cuda_time": 0.16159883620487256, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 57.12, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.2584636356844014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.22680538414718956, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.42, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.326, - "cuda_time_us": 132.73499999999999, - "pct_cuda_time": 1.9599617620297656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.785, - "cuda_time_us": 80.895, - "pct_cuda_time": 1.1944935905330012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.895, - "pct_cuda_time": 1.1944935905330012, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 44.842, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.13372067440344715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.13372067440344715, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.75, - "cuda_time_us": 42.784, - "pct_cuda_time": 0.6317474970933175, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.784, - "pct_cuda_time": 0.6317474970933175, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 905.137, - "cuda_time_us": 196.831, - "pct_cuda_time": 2.9064017296273086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.647, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 638.573, - "cuda_time_us": 57.342999999999996, - "pct_cuda_time": 0.8467253348406438, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.529, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30429722373081264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30429722373081264, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 197.184, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.05575632360285076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.05575632360285076, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 273.177, - "cuda_time_us": 15.199000000000002, - "pct_cuda_time": 0.2244280620867926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.752, - "pct_cuda_time": 0.04063596465970479, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.912, - "pct_cuda_time": 0.16112632498789925, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.535, - "pct_cuda_time": 0.022665772439188537, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.552, - "cuda_time_us": 17.759999999999998, - "pct_cuda_time": 0.26224372542018787, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.23058547388297604, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.571, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.499, - "cuda_time_us": 133.088, - "pct_cuda_time": 1.9651741513920025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.944, - "cuda_time_us": 80.928, - "pct_cuda_time": 1.1949808677255047, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.928, - "pct_cuda_time": 1.1949808677255047, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.229, - "cuda_time_us": 9.152, - "pct_cuda_time": 0.13513820805436708, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.152, - "pct_cuda_time": 0.13513820805436708, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 63.806, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.6350550756121308, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.6350550756121308, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1001.183, - "cuda_time_us": 196.255, - "pct_cuda_time": 2.897896527721789, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.133, - "cuda_time_us": 3.073, - "pct_cuda_time": 0.04537584280496832, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.073, - "pct_cuda_time": 0.04537584280496832, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 647.874, - "cuda_time_us": 56.989999999999995, - "pct_cuda_time": 0.841512945478407, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.4, - "cuda_time_us": 20.639, - "pct_cuda_time": 0.3047549689722555, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.639, - "pct_cuda_time": 0.3047549689722555, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 182.083, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.05859139090469063, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.05859139090469063, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 294.058, - "cuda_time_us": 14.784, - "pct_cuda_time": 0.21830018224166994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.912, - "pct_cuda_time": 0.16112632498789925, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021735515980772332, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.281, - "cuda_time_us": 17.599, - "pct_cuda_time": 0.25986640335979094, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.455, - "pct_cuda_time": 0.22820815182257903, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.502, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 267.93, - "cuda_time_us": 132.96, - "pct_cuda_time": 1.9632841065241096, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.996, - "cuda_time_us": 80.352, - "pct_cuda_time": 1.1864756658199853, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.352, - "pct_cuda_time": 1.1864756658199853, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.621, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.12757802858279413, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.12757802858279413, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 147.407, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6492304121213301, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.6492304121213301, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 917.291, - "cuda_time_us": 195.87, - "pct_cuda_time": 2.8922116271425793, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.863, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048668655348251086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048668655348251086, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 662.168, - "cuda_time_us": 57.662, - "pct_cuda_time": 0.8514356810348466, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.728, - "cuda_time_us": 21.407, - "pct_cuda_time": 0.316095238179615, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.407, - "pct_cuda_time": 0.316095238179615, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 212.089, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.053866278734957515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.053866278734957515, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 283.083, - "cuda_time_us": 15.040000000000001, - "pct_cuda_time": 0.22208027197745647, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.03827340857483824, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.944, - "pct_cuda_time": 0.16159883620487256, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.558, - "cuda_time_us": 17.567, - "pct_cuda_time": 0.25939389214281766, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.22775040658113616, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.03164348556168145, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.908, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.607, - "cuda_time_us": 131.776, - "pct_cuda_time": 1.9458011914960973, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.459, - "cuda_time_us": 79.744, - "pct_cuda_time": 1.1774979526974922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.744, - "pct_cuda_time": 1.1774979526974922, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.024, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1294680734506874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1294680734506874, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.882, - "cuda_time_us": 43.264, - "pct_cuda_time": 0.6388351653479172, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.264, - "pct_cuda_time": 0.6388351653479172, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 872.033, - "cuda_time_us": 196.159, - "pct_cuda_time": 2.8964789940708693, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.021, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045833588046411224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045833588046411224, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 619.152, - "cuda_time_us": 56.543, - "pct_cuda_time": 0.8349125544163111, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.945, - "cuda_time_us": 20.32, - "pct_cuda_time": 0.3000446227780528, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.32, - "pct_cuda_time": 0.3000446227780528, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 194.173, - "cuda_time_us": 3.935, - "pct_cuda_time": 0.0581041137121869, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.935, - "pct_cuda_time": 0.0581041137121869, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 258.631, - "cuda_time_us": 14.911999999999999, - "pct_cuda_time": 0.22019022710956315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.008, - "pct_cuda_time": 0.16254385863881915, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.381, - "cuda_time_us": 17.375999999999998, - "pct_cuda_time": 0.25657359081650816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.2, - "pct_cuda_time": 0.22444282806232296, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03213076275418519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.385, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 165.812, - "cuda_time_us": 133.31199999999998, - "pct_cuda_time": 1.9684817299108155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 56.666, - "cuda_time_us": 80.737, - "pct_cuda_time": 1.1921605663991954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.737, - "pct_cuda_time": 1.1921605663991954, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.934, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.771, - "cuda_time_us": 43.743, - "pct_cuda_time": 0.6459080676269866, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.743, - "pct_cuda_time": 0.6459080676269866, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 923.634, - "cuda_time_us": 196.002, - "pct_cuda_time": 2.8941607359125943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.211, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 657.615, - "cuda_time_us": 57.346, - "pct_cuda_time": 0.8467696327672352, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.098, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.3090223359005458, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.3090223359005458, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 194.754, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05528381238587745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.05528381238587745, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 280.714, - "cuda_time_us": 14.914000000000001, - "pct_cuda_time": 0.22021975906062405, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.433, - "pct_cuda_time": 0.03592561846550209, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.976, - "pct_cuda_time": 0.16207134742184587, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.022222793173276058, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 74.33, - "cuda_time_us": 17.759999999999998, - "pct_cuda_time": 0.26224372542018787, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.648, - "pct_cuda_time": 0.23105798509994938, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.03118574032023856, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.233, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 180.795, - "cuda_time_us": 132.22400000000002, - "pct_cuda_time": 1.9524163485337236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.707, - "cuda_time_us": 80.224, - "pct_cuda_time": 1.1845856209520922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.224, - "pct_cuda_time": 1.1845856209520922, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.801, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13277565196950056, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13277565196950056, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.549, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.6350550756121308, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.6350550756121308, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 901.477, - "cuda_time_us": 196.86100000000002, - "pct_cuda_time": 2.9068447088932214, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.776, - "cuda_time_us": 3.103, - "pct_cuda_time": 0.0458188220708808, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.103, - "pct_cuda_time": 0.0458188220708808, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 640.51, - "cuda_time_us": 57.662, - "pct_cuda_time": 0.8514356810348466, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.971, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.3028796900798927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.3028796900798927, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 192.558, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.05764636847074401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.05764636847074401, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 266.775, - "cuda_time_us": 15.136000000000001, - "pct_cuda_time": 0.2234978056283764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.03827340857483824, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.912, - "pct_cuda_time": 0.16112632498789925, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.02409807206563889, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.962, - "cuda_time_us": 18.11, - "pct_cuda_time": 0.2674118168558335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.999, - "pct_cuda_time": 0.23624084251112537, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.031170974344708148, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.692, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 174.12, - "cuda_time_us": 132.96, - "pct_cuda_time": 1.9632841065241096, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.089, - "cuda_time_us": 80.416, - "pct_cuda_time": 1.187420688253932, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.416, - "pct_cuda_time": 1.187420688253932, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.397, - "cuda_time_us": 9.376, - "pct_cuda_time": 0.1384457865731803, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.376, - "pct_cuda_time": 0.1384457865731803, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.834, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6374176316969973, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6374176316969973, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 909.603, - "cuda_time_us": 196.384, - "pct_cuda_time": 2.899801338565213, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.082, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 644.443, - "cuda_time_us": 56.577, - "pct_cuda_time": 0.8354145975843451, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.852, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.3028796900798927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.3028796900798927, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 199.782, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05433878995193083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05433878995193083, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 272.98, - "cuda_time_us": 14.943999999999999, - "pct_cuda_time": 0.2206627383265365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.04, - "pct_cuda_time": 0.16301636985579246, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 64.145, - "cuda_time_us": 17.441, - "pct_cuda_time": 0.2575333792259852, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.22633287293021623, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.03120050629576898, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.008, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.813, - "cuda_time_us": 133.503, - "pct_cuda_time": 1.9713020312371252, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.981, - "cuda_time_us": 81.567, - "pct_cuda_time": 1.2044163260894407, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.567, - "pct_cuda_time": 1.2044163260894407, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.476, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.13183062953555394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.13183062953555394, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.29, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.6350550756121308, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.008, - "pct_cuda_time": 0.6350550756121308, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 885.181, - "cuda_time_us": 195.712, - "pct_cuda_time": 2.8898786030087735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.875, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 631.825, - "cuda_time_us": 56.832, - "pct_cuda_time": 0.8391799213446013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.687, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.30382471251383936, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.30382471251383936, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 194.339, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.056228834819824075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.056228834819824075, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 265.367, - "cuda_time_us": 14.943999999999999, - "pct_cuda_time": 0.2206627383265365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.04, - "pct_cuda_time": 0.16301636985579246, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.865, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.2584636356844014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.22633287293021623, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03213076275418519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.766, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.161, - "cuda_time_us": 132.512, - "pct_cuda_time": 1.956668949486483, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.612, - "cuda_time_us": 80.735, - "pct_cuda_time": 1.1921310344481346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.735, - "pct_cuda_time": 1.1921310344481346, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.284, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12852305101674075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12852305101674075, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.206, - "cuda_time_us": 43.073, - "pct_cuda_time": 0.6360148640216078, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.073, - "pct_cuda_time": 0.6360148640216078, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 876.389, - "cuda_time_us": 196.353, - "pct_cuda_time": 2.8993435933237706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.601, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 612.0, - "cuda_time_us": 57.249, - "pct_cuda_time": 0.8453373331407847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.753, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.3057147573817326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.3057147573817326, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 175.6, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05433878995193083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05433878995193083, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 266.906, - "cuda_time_us": 14.817, - "pct_cuda_time": 0.21878745943417366, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.913, - "pct_cuda_time": 0.16114109096342966, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 61.911, - "cuda_time_us": 18.048000000000002, - "pct_cuda_time": 0.26649632637294773, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.935, - "pct_cuda_time": 0.23529582007717872, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.03120050629576898, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.216, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.336, - "cuda_time_us": 132.8, - "pct_cuda_time": 1.960921550439243, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.163, - "cuda_time_us": 80.32, - "pct_cuda_time": 1.1860031546030119, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.32, - "pct_cuda_time": 1.1860031546030119, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.576, - "cuda_time_us": 9.632, - "pct_cuda_time": 0.14222587630896677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.632, - "pct_cuda_time": 0.14222587630896677, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.394, - "cuda_time_us": 42.848, - "pct_cuda_time": 0.6326925195272641, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.848, - "pct_cuda_time": 0.6326925195272641, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 993.162, - "cuda_time_us": 196.447, - "pct_cuda_time": 2.900731595023629, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.659, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 730.662, - "cuda_time_us": 57.022000000000006, - "pct_cuda_time": 0.8419854566953804, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.338, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30429722373081264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30429722373081264, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 195.477, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.05764636847074401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.05764636847074401, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 355.597, - "cuda_time_us": 14.88, - "pct_cuda_time": 0.21971771589258987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.035438341272998365, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.976, - "pct_cuda_time": 0.16207134742184587, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.09, - "cuda_time_us": 17.63, - "pct_cuda_time": 0.2603241486012338, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.487, - "pct_cuda_time": 0.22868066303955237, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.03164348556168145, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.92, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.050100954974701434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.050100954974701434, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.263, - "cuda_time_us": 132.832, - "pct_cuda_time": 1.9613940616562162, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.925, - "cuda_time_us": 80.832, - "pct_cuda_time": 1.193563334074585, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.832, - "pct_cuda_time": 1.193563334074585, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.76, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.13041309588463398, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.215, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6374176316969973, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6374176316969973, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 867.633, - "cuda_time_us": 196.96, - "pct_cuda_time": 2.9083065404707327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.324, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04677861048035784, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 615.859, - "cuda_time_us": 57.601, - "pct_cuda_time": 0.8505349565274912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.781, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.3066597798156792, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.3066597798156792, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.93, - "cuda_time_us": 3.713, - "pct_cuda_time": 0.05482606714443456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.713, - "pct_cuda_time": 0.05482606714443456, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 261.026, - "cuda_time_us": 14.976, - "pct_cuda_time": 0.2211352495435098, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.037800897357864925, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.912, - "pct_cuda_time": 0.16112632498789925, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.473, - "cuda_time_us": 18.144, - "pct_cuda_time": 0.26791386002386763, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.0, - "pct_cuda_time": 0.23625560848665578, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.128, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.192, - "cuda_time_us": 132.959, - "pct_cuda_time": 1.963269340548579, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.616, - "cuda_time_us": 80.736, - "pct_cuda_time": 1.1921458004236651, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.736, - "pct_cuda_time": 1.1921458004236651, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.237, - "cuda_time_us": 9.247, - "pct_cuda_time": 0.13654097572975663, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.247, - "pct_cuda_time": 0.13654097572975663, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.77, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6345825643951574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6345825643951574, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 907.179, - "cuda_time_us": 196.99300000000002, - "pct_cuda_time": 2.9087938176632364, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.872, - "cuda_time_us": 3.105, - "pct_cuda_time": 0.04584835402194164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.105, - "pct_cuda_time": 0.04584835402194164, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 648.121, - "cuda_time_us": 56.992000000000004, - "pct_cuda_time": 0.841542477429468, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.533, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.3052422461647593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.3052422461647593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 187.808, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.058118879687717326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.058118879687717326, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 261.649, - "cuda_time_us": 14.848, - "pct_cuda_time": 0.2192452046756166, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.035910852489971674, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.944, - "pct_cuda_time": 0.16159883620487256, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021735515980772332, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 95.931, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2589361469013748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.22680538414718956, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03213076275418519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.552, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.05008618899917102, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.05008618899917102, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.813, - "cuda_time_us": 133.50400000000002, - "pct_cuda_time": 1.971316797212656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.854, - "cuda_time_us": 82.016, - "pct_cuda_time": 1.2110462491025975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.016, - "pct_cuda_time": 1.2110462491025975, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.725, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12994058466766067, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12994058466766067, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.547, - "cuda_time_us": 42.688, - "pct_cuda_time": 0.6303299634423976, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.688, - "pct_cuda_time": 0.6303299634423976, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 909.428, - "cuda_time_us": 196.57500000000002, - "pct_cuda_time": 2.9026216398915228, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.314, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 660.185, - "cuda_time_us": 57.024, - "pct_cuda_time": 0.8420149886464412, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.265, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.3080773134665991, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.3080773134665991, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 197.117, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05481130116890415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05481130116890415, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 282.852, - "cuda_time_us": 14.913, - "pct_cuda_time": 0.22020499308509361, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03638336370694499, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 10.944, - "pct_cuda_time": 0.16159883620487256, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.022222793173276058, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.146, - "cuda_time_us": 17.535, - "pct_cuda_time": 0.2589213809258443, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.392, - "pct_cuda_time": 0.22727789536416285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.03164348556168145, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.88, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04725112169733116, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 167.8, - "cuda_time_us": 133.151, - "pct_cuda_time": 1.9661044078504193, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.703, - "cuda_time_us": 81.087, - "pct_cuda_time": 1.197328657834841, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.087, - "pct_cuda_time": 1.197328657834841, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.055, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12994058466766067, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12994058466766067, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.841, - "cuda_time_us": 43.264, - "pct_cuda_time": 0.6388351653479172, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.264, - "pct_cuda_time": 0.6388351653479172, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 892.037, - "cuda_time_us": 197.374, - "pct_cuda_time": 2.9144196543403247, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.5, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047723632914304474, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 632.276, - "cuda_time_us": 56.704, - "pct_cuda_time": 0.8372898764767082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.278, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.30382471251383936, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.30382471251383936, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 192.84, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.056228834819824075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.056228834819824075, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 265.892, - "cuda_time_us": 14.943999999999999, - "pct_cuda_time": 0.2206627383265365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.035910852489971674, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.008, - "pct_cuda_time": 0.16254385863881915, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.022208027197745644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 9], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.908, - "cuda_time_us": 17.375999999999998, - "pct_cuda_time": 0.25657359081650816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.232, - "pct_cuda_time": 0.2249153392792963, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03165825153721188, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.61, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04630609926338453, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.754, - "cuda_time_us": 134.302, - "pct_cuda_time": 1.9831000456859278, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.638, - "cuda_time_us": 81.439, - "pct_cuda_time": 1.2025262812215474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.439, - "pct_cuda_time": 1.2025262812215474, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.335, - "cuda_time_us": 9.087, - "pct_cuda_time": 0.13417841964489008, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.087, - "pct_cuda_time": 0.13417841964489008, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.983, - "cuda_time_us": 43.776, - "pct_cuda_time": 0.6463953448194902, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.776, - "pct_cuda_time": 0.6463953448194902, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.095, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04629133328785411, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04629133328785411, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 161.25, - "cuda_time_us": 345.951, - "pct_cuda_time": 5.10830400072294, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.008, - "pct_cuda_time": 0.04441605439549129, - "trace": "index_select(bfloat16[2, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010867757990386166, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 342.207, - "pct_cuda_time": 5.053020188337063, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 1181.693, - "cuda_time_us": 119.38999999999999, - "pct_cuda_time": 1.7629098185763643, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.175, - "pct_cuda_time": 0.032115996778654766, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.032603273971158496, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.032603273971158496, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.032603273971158496, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.011340269207359477, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.864, - "pct_cuda_time": 0.012757802858279411, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.767, - "pct_cuda_time": 0.011325503231829062, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.128, - "pct_cuda_time": 0.06095394698955719, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.06804161524415686, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.304, - "pct_cuda_time": 0.50653202459539, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.904, - "pct_cuda_time": 0.4120297812007277, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 1.856, - "pct_cuda_time": 0.027405650584452074, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.736, - "pct_cuda_time": 0.06993166011205011, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 28.16, - "pct_cuda_time": 0.4158098709365142, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.0368558749239183, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - } -} \ No newline at end of file diff --git a/H100_llama8b_pp1_tp1/profiling_bs2_pl1536.json b/H100_llama8b_pp1_tp1/profiling_bs2_pl1536.json deleted file mode 100644 index 6e06cd4b06099fc26211a5f52ccb04ade7ea13bf..0000000000000000000000000000000000000000 --- a/H100_llama8b_pp1_tp1/profiling_bs2_pl1536.json +++ /dev/null @@ -1,18877 +0,0 @@ -{ - "context": { - "python_version": "3.12.9 | packaged by Anaconda, Inc. | (main, Feb 6 2025, 18:56:27) [GCC 11.2.0]", - "torch_version": "2.5.1+cu124", - "engine_args": { - "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "served_model_name": null, - "tokenizer": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "task": "auto", - "skip_tokenizer_init": false, - "tokenizer_mode": "auto", - "trust_remote_code": false, - "allowed_local_media_path": null, - "download_dir": null, - "load_format": "dummy", - "config_format": "auto", - "dtype": "auto", - "kv_cache_dtype": "auto", - "seed": 0, - "max_model_len": null, - "distributed_executor_backend": null, - "pipeline_parallel_size": 1, - "tensor_parallel_size": 1, - "max_parallel_loading_workers": null, - "block_size": null, - "enable_prefix_caching": false, - "disable_sliding_window": false, - "use_v2_block_manager": true, - "swap_space": 4, - "cpu_offload_gb": 0, - "gpu_memory_utilization": 0.9, - "max_num_batched_tokens": 8000, - "max_num_partial_prefills": 1, - "max_long_partial_prefills": 1, - "long_prefill_token_threshold": 0, - "max_num_seqs": 256, - "max_logprobs": 20, - "disable_log_stats": false, - "revision": null, - "code_revision": null, - "rope_scaling": null, - "rope_theta": null, - "hf_overrides": null, - "tokenizer_revision": null, - "quantization": null, - "enforce_eager": true, - "max_seq_len_to_capture": 8192, - "disable_custom_all_reduce": false, - "tokenizer_pool_size": 0, - "tokenizer_pool_type": "ray", - "tokenizer_pool_extra_config": null, - "limit_mm_per_prompt": null, - "mm_processor_kwargs": null, - "disable_mm_preprocessor_cache": false, - "enable_lora": false, - "enable_lora_bias": false, - "max_loras": 1, - "max_lora_rank": 16, - "enable_prompt_adapter": false, - "max_prompt_adapters": 1, - "max_prompt_adapter_token": 0, - "fully_sharded_loras": false, - "lora_extra_vocab_size": 256, - "long_lora_scaling_factors": null, - "lora_dtype": "auto", - "max_cpu_loras": null, - "device": "auto", - "num_scheduler_steps": 1, - "multi_step_stream_outputs": true, - "ray_workers_use_nsight": false, - "num_gpu_blocks_override": null, - "num_lookahead_slots": 0, - "model_loader_extra_config": null, - "ignore_patterns": [], - "preemption_mode": null, - "scheduler_delay_factor": 0.0, - "enable_chunked_prefill": null, - "guided_decoding_backend": "xgrammar", - "logits_processor_pattern": null, - "speculative_model": null, - "speculative_model_quantization": null, - "speculative_draft_tensor_parallel_size": null, - "num_speculative_tokens": null, - "speculative_disable_mqa_scorer": false, - "speculative_max_model_len": null, - "speculative_disable_by_batch_size": null, - "ngram_prompt_lookup_max": null, - "ngram_prompt_lookup_min": null, - "spec_decoding_acceptance_method": "rejection_sampler", - "typical_acceptance_sampler_posterior_threshold": null, - "typical_acceptance_sampler_posterior_alpha": null, - "qlora_adapter_name_or_path": null, - "disable_logprobs_during_spec_decoding": null, - "otlp_traces_endpoint": null, - "collect_detailed_traces": null, - "disable_async_output_proc": false, - "scheduling_policy": "fcfs", - "scheduler_cls": "vllm.core.scheduler.Scheduler", - "override_neuron_config": null, - "override_pooler_config": null, - "compilation_config": null, - "worker_cls": "auto", - "kv_transfer_config": null, - "generation_config": null, - "override_generation_config": null, - "enable_sleep_mode": false, - "model_impl": "auto", - "calculate_kv_scales": false, - "additional_config": null - }, - "prompt_len": 0, - "batch_size": 2, - "num_steps": 2, - "complete_num_requests_per_step": null, - "save_chrome_traces_folder": null - }, - "prefill": { - "metadata": { - "num_running_seqs": null - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 71645.147, - "pct_cuda_time": 99.309817245638, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 95.008, - "pct_cuda_time": 0.13169387616545158, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cuda_time_us": 95.008, - "pct_cuda_time": 0.13169387616545158, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 71515.771, - "pct_cuda_time": 99.13048469550768, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 2133.954000000001, - "pct_cuda_time": 2.957947476199586, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 53.44, - "pct_cuda_time": 0.07407503307386465, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 2080.5140000000006, - "pct_cuda_time": 2.8838724431257203, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 17425.447000000004, - "pct_cuda_time": 24.15401502342582, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 7131.6, - "pct_cuda_time": 9.885357520014468, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 30.462000000000007, - "pct_cuda_time": 0.042224432213623986, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 7101.138, - "pct_cuda_time": 9.843133087800844, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 1362.622, - "pct_cuda_time": 1.8887775021926574, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 1362.622, - "pct_cuda_time": 1.8887775021926574, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 3883.332, - "pct_cuda_time": 5.382820852110721, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 544.833, - "pct_cuda_time": 0.7552118730301814, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 3290.6530000000002, - "pct_cuda_time": 4.561287982964296, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 47.846, - "pct_cuda_time": 0.06632099611624491, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 5047.893, - "pct_cuda_time": 6.997059149107969, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 31.39000000000001, - "pct_cuda_time": 0.0435107651232899, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 5016.503000000001, - "pct_cuda_time": 6.95354838398468, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 51956.37, - "pct_cuda_time": 72.01852219588228, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 32736.348000000005, - "pct_cuda_time": 45.376984670986964, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 32.25300000000001, - "pct_cuda_time": 0.04470699928389516, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 32704.094999999998, - "pct_cuda_time": 45.33227767170306, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 4355.093000000001, - "pct_cuda_time": 6.036745097581521, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 4355.093000000001, - "pct_cuda_time": 6.036745097581521, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 14864.929, - "pct_cuda_time": 20.604792427313807, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 30.656000000000013, - "pct_cuda_time": 0.04249334232620501, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 14834.272999999997, - "pct_cuda_time": 20.562299084987597, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 34.368, - "pct_cuda_time": 0.04763867396486865, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 34.368, - "pct_cuda_time": 0.04763867396486865, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 358.335, - "pct_cuda_time": 0.496700542225361, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.712, - "pct_cuda_time": 0.005145331638663653, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 353.599, - "pct_cuda_time": 0.49013580875534185, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 139.584, - "pct_cuda_time": 0.1934822121366453, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 15.136000000000001, - "pct_cuda_time": 0.02098053331972334, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.672, - "pct_cuda_time": 0.006476020855559424, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 5.312, - "pct_cuda_time": 0.007363147000156607, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 39.36, - "pct_cuda_time": 0.054558257892726655, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 32.576, - "pct_cuda_time": 0.04515472075999654, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0031049415060901355, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 5.216, - "pct_cuda_time": 0.00723007807846703, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 32.768, - "pct_cuda_time": 0.0454208586033757, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.304, - "pct_cuda_time": 0.003193654120549853, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 16913.349, - "cuda_time_us": 71645.147, - "pct_cuda_time": 99.309817245638, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 62.928, - "cuda_time_us": 95.008, - "pct_cuda_time": 0.13169387616545158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 95.008, - "pct_cuda_time": 0.13169387616545158, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[3072]) <- embedding(bfloat16[128256, 4096], int64[3072], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 852.039, - "cuda_time_us": 2282.2039999999997, - "pct_cuda_time": 3.163441930787915, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 50.218, - "cuda_time_us": 53.44, - "pct_cuda_time": 0.07407503307386465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 53.44, - "pct_cuda_time": 0.07407503307386465, - "trace": "_C::rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 624.319, - "cuda_time_us": 559.583, - "pct_cuda_time": 0.7756573583939447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 86.05, - "cuda_time_us": 231.295, - "pct_cuda_time": 0.3206060025228204, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 230.559, - "pct_cuda_time": 0.3195858074565337, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 190.736, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.06019150891091876, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.06019150891091876, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 226.879, - "cuda_time_us": 124.96, - "pct_cuda_time": 0.17321137973259967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.152, - "pct_cuda_time": 0.023774980675204466, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 106.208, - "pct_cuda_time": 0.14721858369590227, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.002217815361492954, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.702, - "cuda_time_us": 159.904, - "pct_cuda_time": 0.2216484672276058, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0010215812008876918, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 159.167, - "pct_cuda_time": 0.22062688602671812, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 23.511, - "cuda_time_us": 32.96, - "pct_cuda_time": 0.04568699644675485, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.96, - "pct_cuda_time": 0.04568699644675485, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 124.43, - "cuda_time_us": 1636.221, - "pct_cuda_time": 2.2680225428733514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 42.657, - "cuda_time_us": 1032.862, - "pct_cuda_time": 1.4316857561889595, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1032.126, - "pct_cuda_time": 1.4306655611226726, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 28.745, - "cuda_time_us": 136.832, - "pct_cuda_time": 0.1896675697148774, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 136.832, - "pct_cuda_time": 0.1896675697148774, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 36.826, - "cuda_time_us": 466.527, - "pct_cuda_time": 0.6466692169695145, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0010188089316858257, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 465.792, - "pct_cuda_time": 0.6456504080378287, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 554.072, - "cuda_time_us": 2255.547, - "pct_cuda_time": 3.126491740730842, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.469, - "cuda_time_us": 33.344, - "pct_cuda_time": 0.04621927213351316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.344, - "pct_cuda_time": 0.04621927213351316, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 398.005, - "cuda_time_us": 551.39, - "pct_cuda_time": 0.7643007576084998, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.755, - "cuda_time_us": 224.671, - "pct_cuda_time": 0.3114242469262396, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 223.647, - "pct_cuda_time": 0.31000484509488413, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 117.229, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.06019150891091876, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.06019150891091876, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 165.874, - "cuda_time_us": 123.232, - "pct_cuda_time": 0.1708161391421873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.152, - "pct_cuda_time": 0.023774980675204466, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 104.672, - "pct_cuda_time": 0.14508948094886903, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0019516775181137992, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 45.444, - "cuda_time_us": 160.063, - "pct_cuda_time": 0.22186886262915415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 158.975, - "pct_cuda_time": 0.22036074818333895, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.62, - "cuda_time_us": 32.736, - "pct_cuda_time": 0.04537650229614583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.736, - "pct_cuda_time": 0.04537650229614583, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.198, - "cuda_time_us": 1638.077, - "pct_cuda_time": 2.2705952086926833, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.694, - "cuda_time_us": 1033.822, - "pct_cuda_time": 1.4330164454058552, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0019516775181137992, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1032.414, - "pct_cuda_time": 1.4310647678877413, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.641, - "cuda_time_us": 137.056, - "pct_cuda_time": 0.18997806386548644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 137.056, - "pct_cuda_time": 0.18997806386548644, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.462, - "cuda_time_us": 467.199, - "pct_cuda_time": 0.6476006994213416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 466.463, - "pct_cuda_time": 0.6465805043550549, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 538.005, - "cuda_time_us": 2257.055, - "pct_cuda_time": 3.1285820317090494, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.665, - "cuda_time_us": 34.528, - "pct_cuda_time": 0.04786045550101794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.528, - "pct_cuda_time": 0.04786045550101794, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 386.651, - "cuda_time_us": 553.122, - "pct_cuda_time": 0.7667015427373158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.922, - "cuda_time_us": 225.76, - "pct_cuda_time": 0.31293374750665576, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0019960338253436584, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 224.32, - "pct_cuda_time": 0.3109377136813121, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 119.131, - "cuda_time_us": 43.553, - "pct_cuda_time": 0.06037032027443913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.553, - "pct_cuda_time": 0.06037032027443913, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.088, - "cuda_time_us": 123.649, - "pct_cuda_time": 0.1713941572707764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.217, - "pct_cuda_time": 0.023865079424265113, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 105.023, - "pct_cuda_time": 0.14557601419379654, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.409, - "pct_cuda_time": 0.0019530636527147324, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.736, - "cuda_time_us": 160.16, - "pct_cuda_time": 0.22200331768544465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 159.392, - "pct_cuda_time": 0.22093876631192805, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.236, - "cuda_time_us": 32.096, - "pct_cuda_time": 0.044489376151548646, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.096, - "pct_cuda_time": 0.044489376151548646, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.687, - "cuda_time_us": 1637.3089999999997, - "pct_cuda_time": 2.2695306573191663, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.457, - "cuda_time_us": 1032.542, - "pct_cuda_time": 1.4312421931166608, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1031.454, - "pct_cuda_time": 1.4297340786708457, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.11, - "cuda_time_us": 137.599, - "pct_cuda_time": 0.19073073495379309, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 137.599, - "pct_cuda_time": 0.19073073495379309, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.632, - "cuda_time_us": 467.16799999999995, - "pct_cuda_time": 0.6475577292487126, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 466.4, - "pct_cuda_time": 0.646493177875196, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 499.273, - "cuda_time_us": 2257.661, - "pct_cuda_time": 3.1294220292772144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.332, - "cuda_time_us": 34.4, - "pct_cuda_time": 0.047683030272098505, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.4, - "pct_cuda_time": 0.047683030272098505, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 357.104, - "cuda_time_us": 552.193, - "pct_cuda_time": 0.7654138236930491, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.716, - "cuda_time_us": 224.577, - "pct_cuda_time": 0.3112939502737519, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0010215812008876918, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 223.84, - "pct_cuda_time": 0.3102723690728642, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.783, - "cuda_time_us": 43.232, - "pct_cuda_time": 0.059925371067539604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.232, - "pct_cuda_time": 0.059925371067539604, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 151.383, - "cuda_time_us": 123.392, - "pct_cuda_time": 0.17103792067833656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.056, - "pct_cuda_time": 0.023641911753514887, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 104.736, - "pct_cuda_time": 0.14517819356332876, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.002217815361492954, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.305, - "cuda_time_us": 160.992, - "pct_cuda_time": 0.22315658167342098, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.0018629649036540812, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 159.648, - "pct_cuda_time": 0.22129361676976692, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.183, - "cuda_time_us": 32.704, - "pct_cuda_time": 0.045332145988915974, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.704, - "pct_cuda_time": 0.045332145988915974, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.59, - "cuda_time_us": 1638.364, - "pct_cuda_time": 2.2709930293231513, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.985, - "cuda_time_us": 1034.431, - "pct_cuda_time": 1.4338606013778237, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0010215812008876918, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1033.694, - "pct_cuda_time": 1.4328390201769359, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.8, - "cuda_time_us": 136.895, - "pct_cuda_time": 0.1897548961947362, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 136.895, - "pct_cuda_time": 0.1897548961947362, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.306, - "cuda_time_us": 467.038, - "pct_cuda_time": 0.6473775317505913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.001861578769053148, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 465.695, - "pct_cuda_time": 0.6455159529815382, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 510.748, - "cuda_time_us": 2252.6010000000006, - "pct_cuda_time": 3.122408188196494, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.048, - "cuda_time_us": 33.441, - "pct_cuda_time": 0.04635372718980367, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.441, - "pct_cuda_time": 0.04635372718980367, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 364.289, - "cuda_time_us": 550.844, - "pct_cuda_time": 0.7635439281163905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.657, - "cuda_time_us": 224.19000000000003, - "pct_cuda_time": 0.31075751618319086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0010188089316858257, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 223.455, - "pct_cuda_time": 0.309738707251505, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.85, - "cuda_time_us": 43.072, - "pct_cuda_time": 0.05970358953139032, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.072, - "pct_cuda_time": 0.05970358953139032, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.362, - "cuda_time_us": 123.83999999999999, - "pct_cuda_time": 0.1716589089795546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.056, - "pct_cuda_time": 0.023641911753514887, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 105.184, - "pct_cuda_time": 0.14579918186454677, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.002217815361492954, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.312, - "cuda_time_us": 159.742, - "pct_cuda_time": 0.2214239134222546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.767, - "pct_cuda_time": 0.0010631652389156848, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 158.975, - "pct_cuda_time": 0.22036074818333895, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.631, - "cuda_time_us": 32.48, - "pct_cuda_time": 0.045021651838306954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.48, - "pct_cuda_time": 0.045021651838306954, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.412, - "cuda_time_us": 1635.8360000000002, - "pct_cuda_time": 2.2674888810519924, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.587, - "cuda_time_us": 1033.053, - "pct_cuda_time": 1.431950507897738, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.055, - "pct_cuda_time": 0.0014623720039844163, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1031.998, - "pct_cuda_time": 1.4304881358937533, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.944, - "cuda_time_us": 136.864, - "pct_cuda_time": 0.18971192602210726, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 136.864, - "pct_cuda_time": 0.18971192602210726, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.218, - "cuda_time_us": 465.919, - "pct_cuda_time": 0.6458264471321471, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 465.151, - "pct_cuda_time": 0.6447618957586306, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 524.908, - "cuda_time_us": 2253.915, - "pct_cuda_time": 3.1242295690621193, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.534, - "cuda_time_us": 33.152, - "pct_cuda_time": 0.045953134290134, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.152, - "pct_cuda_time": 0.045953134290134, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 376.479, - "cuda_time_us": 551.999, - "pct_cuda_time": 0.7651449135804681, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.657, - "cuda_time_us": 224.79999999999998, - "pct_cuda_time": 0.31160305828976, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 224.064, - "pct_cuda_time": 0.3105828632234732, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 114.212, - "cuda_time_us": 42.783, - "pct_cuda_time": 0.05930299663172065, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 42.783, - "pct_cuda_time": 0.05930299663172065, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.588, - "cuda_time_us": 123.80799999999999, - "pct_cuda_time": 0.17161455267232475, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.184, - "pct_cuda_time": 0.023819336982434325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 105.216, - "pct_cuda_time": 0.14584353817177662, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0019516775181137992, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.987, - "cuda_time_us": 160.608, - "pct_cuda_time": 0.2226243059866627, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 159.552, - "pct_cuda_time": 0.22116054784807734, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.154, - "cuda_time_us": 32.384, - "pct_cuda_time": 0.04488858291661738, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.384, - "pct_cuda_time": 0.04488858291661738, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.904, - "cuda_time_us": 1636.38, - "pct_cuda_time": 2.2682429382749, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.627, - "cuda_time_us": 1033.374, - "pct_cuda_time": 1.4323954571046373, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.00190732121088394, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1031.998, - "pct_cuda_time": 1.4304881358937533, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.28, - "cuda_time_us": 136.768, - "pct_cuda_time": 0.1895788571004177, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 136.768, - "pct_cuda_time": 0.1895788571004177, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.694, - "cuda_time_us": 466.238, - "pct_cuda_time": 0.6462686240698449, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.023, - "pct_cuda_time": 0.0014180156967545572, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 465.215, - "pct_cuda_time": 0.6448506083730903, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 501.498, - "cuda_time_us": 2258.335, - "pct_cuda_time": 3.130356283998244, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 22.418, - "cuda_time_us": 33.729, - "pct_cuda_time": 0.04675293395487239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.729, - "pct_cuda_time": 0.04675293395487239, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 353.281, - "cuda_time_us": 551.8720000000001, - "pct_cuda_time": 0.7649688744861497, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.089, - "cuda_time_us": 225.15300000000002, - "pct_cuda_time": 0.31209236380388944, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.281, - "pct_cuda_time": 0.001775638423795296, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 223.872, - "pct_cuda_time": 0.31031672538009414, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.977, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.05983665845307989, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.05983665845307989, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 149.774, - "cuda_time_us": 123.744, - "pct_cuda_time": 0.17152584005786503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.056, - "pct_cuda_time": 0.023641911753514887, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 105.312, - "pct_cuda_time": 0.14597660709346622, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.00190732121088394, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.256, - "cuda_time_us": 159.807, - "pct_cuda_time": 0.22151401217131525, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 159.071, - "pct_cuda_time": 0.22049381710502852, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.16, - "cuda_time_us": 31.775, - "pct_cuda_time": 0.04404442694464913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.775, - "pct_cuda_time": 0.04404442694464913, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.421, - "cuda_time_us": 1640.9589999999998, - "pct_cuda_time": 2.274590048612572, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.094, - "cuda_time_us": 1036.062, - "pct_cuda_time": 1.4361213869119454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1034.974, - "pct_cuda_time": 1.43461327246613, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.036, - "cuda_time_us": 137.376, - "pct_cuda_time": 0.19042162693778503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 137.376, - "pct_cuda_time": 0.19042162693778503, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.006, - "cuda_time_us": 467.521, - "pct_cuda_time": 0.648047034762842, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0010215812008876918, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 466.784, - "pct_cuda_time": 0.6470254535619544, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 540.038, - "cuda_time_us": 2256.826, - "pct_cuda_time": 3.128264606885436, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.293, - "cuda_time_us": 34.24, - "pct_cuda_time": 0.047461248735949216, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.24, - "pct_cuda_time": 0.047461248735949216, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 376.105, - "cuda_time_us": 552.702, - "pct_cuda_time": 0.766119366204924, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.286, - "cuda_time_us": 224.894, - "pct_cuda_time": 0.3117333549422477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.991, - "pct_cuda_time": 0.0013736593895246983, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 223.903, - "pct_cuda_time": 0.31035969555272297, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.994, - "cuda_time_us": 43.072, - "pct_cuda_time": 0.05970358953139032, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.072, - "pct_cuda_time": 0.05970358953139032, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.847, - "cuda_time_us": 124.256, - "pct_cuda_time": 0.1722355409735428, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.024307256361962775, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 105.12, - "pct_cuda_time": 0.14571046925008707, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.002217815361492954, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 46.933, - "cuda_time_us": 160.48000000000002, - "pct_cuda_time": 0.22244688075774327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.0022621716687228127, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 158.848, - "pct_cuda_time": 0.22018470908902044, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.971, - "cuda_time_us": 32.48, - "pct_cuda_time": 0.045021651838306954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.48, - "pct_cuda_time": 0.045021651838306954, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 113.238, - "cuda_time_us": 1637.404, - "pct_cuda_time": 2.2696623401062554, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.269, - "cuda_time_us": 1033.373, - "pct_cuda_time": 1.4323940709700362, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.695, - "pct_cuda_time": 0.002349498148581598, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1031.678, - "pct_cuda_time": 1.4300445728214548, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.645, - "cuda_time_us": 137.568, - "pct_cuda_time": 0.19068776478116417, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 137.568, - "pct_cuda_time": 0.19068776478116417, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 41.395, - "cuda_time_us": 466.46299999999997, - "pct_cuda_time": 0.6465805043550548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 465.407, - "pct_cuda_time": 0.6451167462164694, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 517.167, - "cuda_time_us": 2252.4130000000005, - "pct_cuda_time": 3.1221475948915183, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.339, - "cuda_time_us": 34.976, - "pct_cuda_time": 0.048481443802235964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.976, - "pct_cuda_time": 0.048481443802235964, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 367.338, - "cuda_time_us": 551.072, - "pct_cuda_time": 0.7638599668054031, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.443, - "cuda_time_us": 224.256, - "pct_cuda_time": 0.3108490010668524, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 223.52, - "pct_cuda_time": 0.30982880600056567, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.608, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.0600584399892292, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.0600584399892292, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.676, - "cuda_time_us": 123.74399999999999, - "pct_cuda_time": 0.17152584005786503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.184, - "pct_cuda_time": 0.023819336982434325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 104.832, - "pct_cuda_time": 0.14531126248501833, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.728, - "pct_cuda_time": 0.0023952405904123903, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.781, - "cuda_time_us": 159.744, - "pct_cuda_time": 0.2214266856914565, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 159.008, - "pct_cuda_time": 0.2204064906251698, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.907, - "cuda_time_us": 32.384, - "pct_cuda_time": 0.04488858291661738, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.384, - "pct_cuda_time": 0.04488858291661738, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.445, - "cuda_time_us": 1633.9810000000002, - "pct_cuda_time": 2.2649176013672614, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.827, - "cuda_time_us": 1031.806, - "pct_cuda_time": 1.4302219980503743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1030.75, - "pct_cuda_time": 1.4287582399117889, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.51, - "cuda_time_us": 136.735, - "pct_cuda_time": 0.1895331146585869, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 136.735, - "pct_cuda_time": 0.1895331146585869, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 35.993, - "cuda_time_us": 465.44, - "pct_cuda_time": 0.6451624886583003, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0010215812008876918, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 464.703, - "pct_cuda_time": 0.6441409074574125, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 494.265, - "cuda_time_us": 2253.753, - "pct_cuda_time": 3.124005015256768, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.962, - "cuda_time_us": 34.015, - "pct_cuda_time": 0.04714936845073926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.015, - "pct_cuda_time": 0.04714936845073926, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 343.206, - "cuda_time_us": 551.55, - "pct_cuda_time": 0.7645225391446491, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.729, - "cuda_time_us": 225.18200000000002, - "pct_cuda_time": 0.31213256170731646, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.055, - "pct_cuda_time": 0.0014623720039844163, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 224.127, - "pct_cuda_time": 0.31067018970333204, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.183, - "cuda_time_us": 42.912, - "pct_cuda_time": 0.05948180799524101, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 42.912, - "pct_cuda_time": 0.05948180799524101, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 147.541, - "cuda_time_us": 123.456, - "pct_cuda_time": 0.17112663329279632, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.96, - "pct_cuda_time": 0.02350884283182531, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 105.088, - "pct_cuda_time": 0.1456661129428572, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0019516775181137992, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.353, - "cuda_time_us": 160.0, - "pct_cuda_time": 0.22178153614929538, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 159.264, - "pct_cuda_time": 0.22076134108300863, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.366, - "cuda_time_us": 31.329, - "pct_cuda_time": 0.04342621091263297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.329, - "pct_cuda_time": 0.04342621091263297, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 105.821, - "cuda_time_us": 1636.8590000000002, - "pct_cuda_time": 2.2689068967487467, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.443, - "cuda_time_us": 1032.989, - "pct_cuda_time": 1.431861795283278, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.375, - "pct_cuda_time": 0.0019059350762830071, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1031.614, - "pct_cuda_time": 1.429955860206995, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.318, - "cuda_time_us": 137.248, - "pct_cuda_time": 0.19024420170886555, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 137.248, - "pct_cuda_time": 0.19024420170886555, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.074, - "cuda_time_us": 466.622, - "pct_cuda_time": 0.6468008997566032, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.055, - "pct_cuda_time": 0.0014623720039844163, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 465.567, - "pct_cuda_time": 0.6453385277526188, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 499.068, - "cuda_time_us": 2253.7899999999995, - "pct_cuda_time": 3.1240563022370016, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.864, - "cuda_time_us": 34.209, - "pct_cuda_time": 0.04741827856332029, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.209, - "pct_cuda_time": 0.04741827856332029, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 357.814, - "cuda_time_us": 551.4879999999999, - "pct_cuda_time": 0.7644365987993912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.484, - "cuda_time_us": 224.289, - "pct_cuda_time": 0.3108947435086831, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0010215812008876918, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 223.552, - "pct_cuda_time": 0.3098731623077955, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.589, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.0605907156759875, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.0605907156759875, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.234, - "cuda_time_us": 123.392, - "pct_cuda_time": 0.17103792067833656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.216, - "pct_cuda_time": 0.023863693289664183, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 104.736, - "pct_cuda_time": 0.14517819356332876, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0019960338253436584, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.753, - "cuda_time_us": 160.095, - "pct_cuda_time": 0.22191321893638402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 159.359, - "pct_cuda_time": 0.22089302387009727, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.941, - "cuda_time_us": 31.679, - "pct_cuda_time": 0.04391135802295955, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.679, - "pct_cuda_time": 0.04391135802295955, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.145, - "cuda_time_us": 1636.4139999999998, - "pct_cuda_time": 2.2682900668513315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.917, - "cuda_time_us": 1032.991, - "pct_cuda_time": 1.43186456755248, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0014651442731862824, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1031.934, - "pct_cuda_time": 1.4303994232792934, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.455, - "cuda_time_us": 137.12, - "pct_cuda_time": 0.19006677647994613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 137.12, - "pct_cuda_time": 0.19006677647994613, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.445, - "cuda_time_us": 466.303, - "pct_cuda_time": 0.6463587228189055, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 465.567, - "pct_cuda_time": 0.6453385277526188, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 505.42, - "cuda_time_us": 2182.367, - "pct_cuda_time": 3.0250544106345583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.997, - "cuda_time_us": 33.856, - "pct_cuda_time": 0.04692897304919091, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.856, - "pct_cuda_time": 0.04692897304919091, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 355.679, - "cuda_time_us": 550.7529999999999, - "pct_cuda_time": 0.7634177898677054, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.616, - "cuda_time_us": 224.864, - "pct_cuda_time": 0.31169177090421973, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 223.776, - "pct_cuda_time": 0.3101836564584045, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.655, - "cuda_time_us": 42.913, - "pct_cuda_time": 0.05948319412984195, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 42.913, - "pct_cuda_time": 0.05948319412984195, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 149.269, - "cuda_time_us": 123.10399999999998, - "pct_cuda_time": 0.17063871391326785, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.895, - "pct_cuda_time": 0.023418744082764658, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 104.609, - "pct_cuda_time": 0.14500215446901024, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.002217815361492954, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.089, - "cuda_time_us": 159.872, - "pct_cuda_time": 0.22160411092037596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0015524707530450677, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 158.752, - "pct_cuda_time": 0.22005164016733086, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.222, - "cuda_time_us": 31.648, - "pct_cuda_time": 0.04386838785033062, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.648, - "pct_cuda_time": 0.04386838785033062, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.53, - "cuda_time_us": 1566.1100000000001, - "pct_cuda_time": 2.170839259867331, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 42.075, - "cuda_time_us": 990.398, - "pct_cuda_time": 1.3728249364949365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 989.662, - "pct_cuda_time": 1.3718047414286498, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.023, - "cuda_time_us": 133.12, - "pct_cuda_time": 0.18452223807621376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 133.12, - "pct_cuda_time": 0.18452223807621376, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.317, - "cuda_time_us": 442.59200000000004, - "pct_cuda_time": 0.6134920852961809, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.345, - "pct_cuda_time": 0.0018643510382550142, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 441.247, - "pct_cuda_time": 0.6116277342579259, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 489.931, - "cuda_time_us": 2106.5860000000002, - "pct_cuda_time": 2.9200117444412474, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.738, - "cuda_time_us": 33.312, - "pct_cuda_time": 0.04617491582628329, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.312, - "pct_cuda_time": 0.04617491582628329, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 354.735, - "cuda_time_us": 518.335, - "pct_cuda_time": 0.7184820783746564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.509, - "cuda_time_us": 211.487, - "pct_cuda_time": 0.29314944834753764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 210.719, - "pct_cuda_time": 0.29208489697402107, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 94.17, - "cuda_time_us": 41.696, - "pct_cuda_time": 0.05779626832050637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 41.696, - "pct_cuda_time": 0.05779626832050637, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.692, - "cuda_time_us": 115.55199999999999, - "pct_cuda_time": 0.16017062540702112, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.672, - "pct_cuda_time": 0.02310963606675658, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 97.216, - "pct_cuda_time": 0.13475446136431185, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.002306527975952672, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.452, - "cuda_time_us": 149.60000000000002, - "pct_cuda_time": 0.2073657362995912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.544, - "pct_cuda_time": 0.20590197816100583, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.571, - "cuda_time_us": 32.801, - "pct_cuda_time": 0.04546660104520649, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.801, - "pct_cuda_time": 0.04546660104520649, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 91.274, - "cuda_time_us": 1522.138, - "pct_cuda_time": 2.109888149195101, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 31.488, - "cuda_time_us": 951.646, - "pct_cuda_time": 1.3191094484395771, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 950.91, - "pct_cuda_time": 1.3180892533732902, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 19.857, - "cuda_time_us": 131.423, - "pct_cuda_time": 0.1821699676584303, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 131.423, - "pct_cuda_time": 0.1821699676584303, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.215, - "cuda_time_us": 439.069, - "pct_cuda_time": 0.6086087330970936, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0010188089316858257, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 438.334, - "pct_cuda_time": 0.6075899241654077, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 510.703, - "cuda_time_us": 2094.013, - "pct_cuda_time": 2.9025838741037155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.164, - "cuda_time_us": 33.44, - "pct_cuda_time": 0.04635234105520273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.44, - "pct_cuda_time": 0.04635234105520273, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 356.247, - "cuda_time_us": 505.76, - "pct_cuda_time": 0.7010514357679226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.553, - "cuda_time_us": 209.184, - "pct_cuda_time": 0.28995718036158874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.0013750455241256314, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.192, - "pct_cuda_time": 0.28858213483746314, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.538, - "cuda_time_us": 40.703, - "pct_cuda_time": 0.05641983666177981, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.703, - "pct_cuda_time": 0.05641983666177981, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 149.924, - "cuda_time_us": 110.657, - "pct_cuda_time": 0.1533854965354536, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.16, - "pct_cuda_time": 0.022399935151078833, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 93.184, - "pct_cuda_time": 0.12916556665334963, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.313, - "pct_cuda_time": 0.001819994731025155, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.575, - "cuda_time_us": 145.216, - "pct_cuda_time": 0.20128892220910047, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.704, - "pct_cuda_time": 0.0009758387590568996, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 144.512, - "pct_cuda_time": 0.2003130834500436, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.615, - "cuda_time_us": 32.032, - "pct_cuda_time": 0.04440066353708893, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.032, - "pct_cuda_time": 0.04440066353708893, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.506, - "cuda_time_us": 1522.781, - "pct_cuda_time": 2.110779433743501, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.828, - "cuda_time_us": 950.974, - "pct_cuda_time": 1.3181779659877502, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0019960338253436584, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 949.534, - "pct_cuda_time": 1.3161819321624064, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.656, - "cuda_time_us": 132.128, - "pct_cuda_time": 0.18314719255208808, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 132.128, - "pct_cuda_time": 0.18314719255208808, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.139, - "cuda_time_us": 439.679, - "pct_cuda_time": 0.6094542752036627, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 438.623, - "pct_cuda_time": 0.6079905170650773, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 480.564, - "cuda_time_us": 2092.414, - "pct_cuda_time": 2.9003674448768235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.905, - "cuda_time_us": 33.248, - "pct_cuda_time": 0.04608620321182357, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.248, - "pct_cuda_time": 0.04608620321182357, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 340.462, - "cuda_time_us": 505.98400000000004, - "pct_cuda_time": 0.7013619299185317, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.156, - "cuda_time_us": 208.544, - "pct_cuda_time": 0.2890700542169916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.001818608596424222, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 207.232, - "pct_cuda_time": 0.28725144562056737, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 93.552, - "cuda_time_us": 40.864, - "pct_cuda_time": 0.05664300433253003, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.864, - "pct_cuda_time": 0.05664300433253003, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 148.397, - "cuda_time_us": 110.752, - "pct_cuda_time": 0.15351717932254225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.02248864776553855, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 93.184, - "pct_cuda_time": 0.12916556665334963, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.0018629649036540812, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.252, - "cuda_time_us": 145.82399999999998, - "pct_cuda_time": 0.2021316920464678, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.248, - "pct_cuda_time": 0.0017298959819645038, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 144.576, - "pct_cuda_time": 0.2004017960645033, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.701, - "cuda_time_us": 31.04, - "pct_cuda_time": 0.0430256180129633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.04, - "pct_cuda_time": 0.0430256180129633, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.187, - "cuda_time_us": 1522.142, - "pct_cuda_time": 2.109893693733505, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.817, - "cuda_time_us": 950.5260000000001, - "pct_cuda_time": 1.3175569776865321, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 949.47, - "pct_cuda_time": 1.316093219547947, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.67, - "cuda_time_us": 131.584, - "pct_cuda_time": 0.1823931353291805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 131.584, - "pct_cuda_time": 0.1823931353291805, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.098, - "cuda_time_us": 440.032, - "pct_cuda_time": 0.6099435807177921, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 439.008, - "pct_cuda_time": 0.6085241788864366, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 516.452, - "cuda_time_us": 2094.907, - "pct_cuda_time": 2.90382307843695, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.32, - "cuda_time_us": 34.464, - "pct_cuda_time": 0.04777174288655822, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.464, - "pct_cuda_time": 0.04777174288655822, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 357.867, - "cuda_time_us": 507.29400000000004, - "pct_cuda_time": 0.703177766245754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.52, - "cuda_time_us": 209.50300000000001, - "pct_cuda_time": 0.29039935729928645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0013306892168957723, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.543, - "pct_cuda_time": 0.2890686680823907, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.568, - "cuda_time_us": 40.448, - "pct_cuda_time": 0.05606637233854187, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.448, - "pct_cuda_time": 0.05606637233854187, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.334, - "cuda_time_us": 110.62400000000001, - "pct_cuda_time": 0.15333975409362283, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.02248864776553855, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 92.896, - "pct_cuda_time": 0.1287663598882809, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0020847464398033766, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.593, - "cuda_time_us": 146.719, - "pct_cuda_time": 0.2033722825143029, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0019516775181137992, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.311, - "pct_cuda_time": 0.2014206049961891, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.086, - "cuda_time_us": 30.751, - "pct_cuda_time": 0.042625025113293635, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 30.751, - "pct_cuda_time": 0.042625025113293635, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.07, - "cuda_time_us": 1522.3980000000001, - "pct_cuda_time": 2.110248544191344, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.392, - "cuda_time_us": 951.2950000000001, - "pct_cuda_time": 1.3186229151946498, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 950.527, - "pct_cuda_time": 1.317558363821133, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.005, - "cuda_time_us": 131.743, - "pct_cuda_time": 0.1826135307307289, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 131.743, - "pct_cuda_time": 0.1826135307307289, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.803, - "cuda_time_us": 439.36, - "pct_cuda_time": 0.609012098265965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0014651442731862824, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 438.303, - "pct_cuda_time": 0.6075469539927788, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 502.483, - "cuda_time_us": 2092.766, - "pct_cuda_time": 2.9008553642563517, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.225, - "cuda_time_us": 34.017, - "pct_cuda_time": 0.04715214071994113, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.017, - "pct_cuda_time": 0.04715214071994113, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 359.536, - "cuda_time_us": 505.28, - "pct_cuda_time": 0.7003860911594747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.776, - "cuda_time_us": 208.959, - "pct_cuda_time": 0.2896453000763788, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0010188089316858257, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.224, - "pct_cuda_time": 0.288626491144693, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.888, - "cuda_time_us": 40.128, - "pct_cuda_time": 0.05562280926624328, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.128, - "pct_cuda_time": 0.05562280926624328, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.659, - "cuda_time_us": 111.041, - "pct_cuda_time": 0.15391777222221192, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.385, - "pct_cuda_time": 0.022711815436288783, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 93.151, - "pct_cuda_time": 0.12911982421151882, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.0020861325744043094, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.878, - "cuda_time_us": 145.15200000000002, - "pct_cuda_time": 0.20120020959464077, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 144.096, - "pct_cuda_time": 0.1997364514560554, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.057, - "cuda_time_us": 31.424, - "pct_cuda_time": 0.04355789369972161, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.424, - "pct_cuda_time": 0.04355789369972161, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.608, - "cuda_time_us": 1522.045, - "pct_cuda_time": 2.1097592386772144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.871, - "cuda_time_us": 951.6460000000001, - "pct_cuda_time": 1.3191094484395771, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 950.878, - "pct_cuda_time": 1.3180448970660608, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.053, - "cuda_time_us": 131.712, - "pct_cuda_time": 0.18257056055809995, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 131.712, - "pct_cuda_time": 0.18257056055809995, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.654, - "cuda_time_us": 438.68699999999995, - "pct_cuda_time": 0.608079229679537, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 437.919, - "pct_cuda_time": 0.6070146783060204, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 478.806, - "cuda_time_us": 2094.9089999999997, - "pct_cuda_time": 2.9038258507061507, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.254, - "cuda_time_us": 32.64, - "pct_cuda_time": 0.04524343337445626, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.64, - "pct_cuda_time": 0.04524343337445626, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 331.709, - "cuda_time_us": 506.75199999999995, - "pct_cuda_time": 0.7024264812920482, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.547, - "cuda_time_us": 209.92, - "pct_cuda_time": 0.2909773754278755, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.896, - "pct_cuda_time": 0.28955797359652, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 91.406, - "cuda_time_us": 40.192, - "pct_cuda_time": 0.055711521880703, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.192, - "pct_cuda_time": 0.055711521880703, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 142.989, - "cuda_time_us": 111.424, - "pct_cuda_time": 0.1544486617743693, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.576, - "pct_cuda_time": 0.022976567145067, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 93.504, - "pct_cuda_time": 0.12960912972564823, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.0018629649036540812, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.452, - "cuda_time_us": 145.21599999999998, - "pct_cuda_time": 0.20128892220910044, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 144.48, - "pct_cuda_time": 0.20026872714281369, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 22.882, - "cuda_time_us": 31.776, - "pct_cuda_time": 0.04404581307925006, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.776, - "pct_cuda_time": 0.04404581307925006, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.236, - "cuda_time_us": 1523.741, - "pct_cuda_time": 2.112110122960397, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.624, - "cuda_time_us": 952.6060000000001, - "pct_cuda_time": 1.320440137656473, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0019960338253436584, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 951.166, - "pct_cuda_time": 1.3184441038311294, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.024, - "cuda_time_us": 131.68, - "pct_cuda_time": 0.1825262042508701, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 131.68, - "pct_cuda_time": 0.1825262042508701, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.989, - "cuda_time_us": 439.455, - "pct_cuda_time": 0.6091437810530538, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 438.719, - "pct_cuda_time": 0.608123585986767, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 489.608, - "cuda_time_us": 2097.725, - "pct_cuda_time": 2.907729205742379, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.0, - "cuda_time_us": 34.016, - "pct_cuda_time": 0.047150754585340196, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.016, - "pct_cuda_time": 0.047150754585340196, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 346.716, - "cuda_time_us": 507.552, - "pct_cuda_time": 0.7035353889727948, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.219, - "cuda_time_us": 209.63199999999998, - "pct_cuda_time": 0.2905781686628068, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.896, - "pct_cuda_time": 0.28955797359652, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.922, - "cuda_time_us": 40.864, - "pct_cuda_time": 0.05664300433253003, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.864, - "pct_cuda_time": 0.05664300433253003, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 148.154, - "cuda_time_us": 110.656, - "pct_cuda_time": 0.15338411040085267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.448, - "pct_cuda_time": 0.022799141916147563, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 92.832, - "pct_cuda_time": 0.12867764727382117, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.00190732121088394, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.853, - "cuda_time_us": 146.4, - "pct_cuda_time": 0.20293010557660526, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.376, - "pct_cuda_time": 0.20151070374524976, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.458, - "cuda_time_us": 32.097, - "pct_cuda_time": 0.044490762286149586, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.097, - "pct_cuda_time": 0.044490762286149586, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.724, - "cuda_time_us": 1524.06, - "pct_cuda_time": 2.1125522998980943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.839, - "cuda_time_us": 953.246, - "pct_cuda_time": 1.3213272638010702, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 952.51, - "pct_cuda_time": 1.3203070687347833, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.738, - "cuda_time_us": 131.583, - "pct_cuda_time": 0.18239174919457957, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 131.583, - "pct_cuda_time": 0.18239174919457957, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.562, - "cuda_time_us": 439.231, - "pct_cuda_time": 0.6088332869024448, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.153, - "pct_cuda_time": 0.0015982131948758598, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 438.078, - "pct_cuda_time": 0.6072350737075688, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 508.756, - "cuda_time_us": 2095.5789999999997, - "pct_cuda_time": 2.904754560888776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.265, - "cuda_time_us": 33.695, - "pct_cuda_time": 0.04670580537844067, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.695, - "pct_cuda_time": 0.04670580537844067, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 356.449, - "cuda_time_us": 506.814, - "pct_cuda_time": 0.7025124216373062, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.861, - "cuda_time_us": 209.311, - "pct_cuda_time": 0.29013321945590725, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0013306892168957723, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.351, - "pct_cuda_time": 0.2888025302390115, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 93.485, - "cuda_time_us": 40.864, - "pct_cuda_time": 0.05664300433253003, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.864, - "pct_cuda_time": 0.05664300433253003, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.051, - "cuda_time_us": 110.848, - "pct_cuda_time": 0.15365024824423182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.064, - "pct_cuda_time": 0.022266866229389254, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 93.28, - "pct_cuda_time": 0.1292986355750392, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0020847464398033766, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.672, - "cuda_time_us": 145.791, - "pct_cuda_time": 0.202085949604637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.00190732121088394, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 144.415, - "pct_cuda_time": 0.20017862839375308, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.318, - "cuda_time_us": 31.457, - "pct_cuda_time": 0.0436036361415524, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.457, - "pct_cuda_time": 0.0436036361415524, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.551, - "cuda_time_us": 1523.6129999999998, - "pct_cuda_time": 2.111932697731477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.945, - "cuda_time_us": 950.846, - "pct_cuda_time": 1.3180005407588307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 949.758, - "pct_cuda_time": 1.3164924263130156, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.786, - "cuda_time_us": 132.416, - "pct_cuda_time": 0.18354639931715686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 132.416, - "pct_cuda_time": 0.18354639931715686, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.932, - "cuda_time_us": 440.351, - "pct_cuda_time": 0.6103857576554899, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0019516775181137992, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 438.943, - "pct_cuda_time": 0.608434080137376, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 498.435, - "cuda_time_us": 2094.205, - "pct_cuda_time": 2.9028500119470944, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.96, - "cuda_time_us": 33.024, - "pct_cuda_time": 0.045775709061214566, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.024, - "pct_cuda_time": 0.045775709061214566, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 356.593, - "cuda_time_us": 507.008, - "pct_cuda_time": 0.7027813317498872, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.482, - "cuda_time_us": 209.375, - "pct_cuda_time": 0.290221932070367, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0013306892168957723, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.415, - "pct_cuda_time": 0.28889124285347123, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.114, - "cuda_time_us": 40.576, - "pct_cuda_time": 0.05624379756746131, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.576, - "pct_cuda_time": 0.05624379756746131, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.164, - "cuda_time_us": 111.105, - "pct_cuda_time": 0.15400648483667165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.289, - "pct_cuda_time": 0.022578746514599204, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 93.312, - "pct_cuda_time": 0.12934299188226905, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0020847464398033766, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.591, - "cuda_time_us": 145.952, - "pct_cuda_time": 0.20230911727538722, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 144.928, - "pct_cuda_time": 0.20088971544403172, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.15, - "cuda_time_us": 31.712, - "pct_cuda_time": 0.043957100464790344, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.712, - "pct_cuda_time": 0.043957100464790344, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.03, - "cuda_time_us": 1522.461, - "pct_cuda_time": 2.1103358706712023, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.734, - "cuda_time_us": 951.87, - "pct_cuda_time": 1.319419942590186, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 951.134, - "pct_cuda_time": 1.3183997475238993, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.683, - "cuda_time_us": 131.744, - "pct_cuda_time": 0.18261491686532982, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 131.744, - "pct_cuda_time": 0.18261491686532982, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.951, - "cuda_time_us": 438.84700000000004, - "pct_cuda_time": 0.6083010112156865, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 437.759, - "pct_cuda_time": 0.6067928967698712, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 483.889, - "cuda_time_us": 2189.3999999999996, - "pct_cuda_time": 3.03480309528292, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.094, - "cuda_time_us": 33.728, - "pct_cuda_time": 0.046751547820271466, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.728, - "pct_cuda_time": 0.046751547820271466, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 344.999, - "cuda_time_us": 506.942, - "pct_cuda_time": 0.7026898468662256, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.154, - "cuda_time_us": 209.27800000000002, - "pct_cuda_time": 0.2900874770140765, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0010188089316858257, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.543, - "pct_cuda_time": 0.2890686680823907, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 94.459, - "cuda_time_us": 40.32, - "pct_cuda_time": 0.05588894710962243, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 40.32, - "pct_cuda_time": 0.05588894710962243, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 145.282, - "cuda_time_us": 111.296, - "pct_cuda_time": 0.15427123654544986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 16.192, - "pct_cuda_time": 0.02244429145830869, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 93.792, - "pct_cuda_time": 0.13000833649071694, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.001818608596424222, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.716, - "cuda_time_us": 146.048, - "pct_cuda_time": 0.2024421861970768, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 145.312, - "pct_cuda_time": 0.2014219911307901, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.886, - "cuda_time_us": 31.391, - "pct_cuda_time": 0.04351215125789081, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.391, - "pct_cuda_time": 0.04351215125789081, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.594, - "cuda_time_us": 1617.339, - "pct_cuda_time": 2.2418495493385326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.671, - "cuda_time_us": 1009.053, - "pct_cuda_time": 1.3986832774753435, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.375, - "pct_cuda_time": 0.0019059350762830071, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1007.678, - "pct_cuda_time": 1.3967773423990604, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 19.783, - "cuda_time_us": 137.855, - "pct_cuda_time": 0.19108558541163195, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 137.855, - "pct_cuda_time": 0.19108558541163195, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.547, - "cuda_time_us": 470.431, - "pct_cuda_time": 0.6520806864515574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 469.663, - "pct_cuda_time": 0.6510161350780407, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 506.49, - "cuda_time_us": 2332.508, - "pct_cuda_time": 3.233170045753254, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.732, - "cuda_time_us": 34.721, - "pct_cuda_time": 0.04812797947899802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.721, - "pct_cuda_time": 0.04812797947899802, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 358.153, - "cuda_time_us": 555.1030000000001, - "pct_cuda_time": 0.7694474753817645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.224, - "cuda_time_us": 226.174, - "pct_cuda_time": 0.31350760723144205, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.535, - "pct_cuda_time": 0.0021277166124323023, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 224.639, - "pct_cuda_time": 0.3113798906190098, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.428, - "cuda_time_us": 42.88, - "pct_cuda_time": 0.05943745168801117, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 42.88, - "pct_cuda_time": 0.05943745168801117, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 150.152, - "cuda_time_us": 125.02400000000002, - "pct_cuda_time": 0.17330009234705943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.024218543747503055, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 106.048, - "pct_cuda_time": 0.146996802159753, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.0020847464398033766, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 45.149, - "cuda_time_us": 161.025, - "pct_cuda_time": 0.22320232411525182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.025, - "pct_cuda_time": 0.0014207879659564233, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 160.0, - "pct_cuda_time": 0.22178153614929538, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.076, - "cuda_time_us": 33.024, - "pct_cuda_time": 0.045775709061214566, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.024, - "pct_cuda_time": 0.045775709061214566, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.753, - "cuda_time_us": 1709.6599999999999, - "pct_cuda_time": 2.369818881831277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.682, - "cuda_time_us": 1082.366, - "pct_cuda_time": 1.5003049634735515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1081.598, - "pct_cuda_time": 1.4992404121000347, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.127, - "cuda_time_us": 138.655, - "pct_cuda_time": 0.19219449309237843, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 138.655, - "pct_cuda_time": 0.19219449309237843, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.157, - "cuda_time_us": 488.639, - "pct_cuda_time": 0.6773194252653472, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.152, - "pct_cuda_time": 0.0015968270602749264, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 487.487, - "pct_cuda_time": 0.6757225982050722, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 517.637, - "cuda_time_us": 2353.8509999999997, - "pct_cuda_time": 3.2627543165409687, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.227, - "cuda_time_us": 33.76, - "pct_cuda_time": 0.04679590412750132, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.76, - "pct_cuda_time": 0.04679590412750132, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 367.185, - "cuda_time_us": 576.1279999999999, - "pct_cuda_time": 0.7985909553663826, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.981, - "cuda_time_us": 234.368, - "pct_cuda_time": 0.32486559415148786, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 233.632, - "pct_cuda_time": 0.3238453990852011, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.847, - "cuda_time_us": 44.448, - "pct_cuda_time": 0.061610910742274254, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.448, - "pct_cuda_time": 0.061610910742274254, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.932, - "cuda_time_us": 128.929, - "pct_cuda_time": 0.17871294796370313, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.6, - "pct_cuda_time": 0.024395968976422492, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 109.696, - "pct_cuda_time": 0.15205342118395693, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.633, - "pct_cuda_time": 0.002263557803323746, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.003, - "cuda_time_us": 168.38299999999998, - "pct_cuda_time": 0.23340150250891747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.001861578769053148, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 167.04, - "pct_cuda_time": 0.23153992373986435, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.789, - "cuda_time_us": 32.0, - "pct_cuda_time": 0.044356307229859074, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.0, - "pct_cuda_time": 0.044356307229859074, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.842, - "cuda_time_us": 1711.963, - "pct_cuda_time": 2.3730111498172257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.712, - "cuda_time_us": 1084.542, - "pct_cuda_time": 1.5033211923651817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1083.518, - "pct_cuda_time": 1.5019017905338263, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.178, - "cuda_time_us": 139.487, - "pct_cuda_time": 0.19334775708035476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 139.487, - "pct_cuda_time": 0.19334775708035476, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.291, - "cuda_time_us": 487.93399999999997, - "pct_cuda_time": 0.6763422003716892, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 487.198, - "pct_cuda_time": 0.6753220053054025, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 533.165, - "cuda_time_us": 2355.262, - "pct_cuda_time": 3.264710152462886, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.428, - "cuda_time_us": 34.432, - "pct_cuda_time": 0.04772738657932837, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.432, - "pct_cuda_time": 0.04772738657932837, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 375.446, - "cuda_time_us": 574.817, - "pct_cuda_time": 0.7967737329045596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.999, - "cuda_time_us": 235.201, - "pct_cuda_time": 0.3260202442740651, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.025, - "pct_cuda_time": 0.0014207879659564233, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 234.176, - "pct_cuda_time": 0.32459945630810866, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.538, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.06085685351936666, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.06085685351936666, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.462, - "cuda_time_us": 128.576, - "pct_cuda_time": 0.17822364244957375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.344, - "pct_cuda_time": 0.02404111851858362, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 109.632, - "pct_cuda_time": 0.1519647085694972, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.002217815361492954, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 46.724, - "cuda_time_us": 167.136, - "pct_cuda_time": 0.23167299266155394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 166.4, - "pct_cuda_time": 0.2306527975952672, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 21.36, - "cuda_time_us": 32.64, - "pct_cuda_time": 0.04524343337445626, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.64, - "pct_cuda_time": 0.04524343337445626, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 103.729, - "cuda_time_us": 1713.373, - "pct_cuda_time": 2.3749655996045416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.813, - "cuda_time_us": 1084.67, - "pct_cuda_time": 1.5034986175941014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1083.934, - "pct_cuda_time": 1.5024784225278145, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.74, - "cuda_time_us": 139.104, - "pct_cuda_time": 0.1928168675281974, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 139.104, - "pct_cuda_time": 0.1928168675281974, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.493, - "cuda_time_us": 489.59900000000005, - "pct_cuda_time": 0.678650114482243, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 488.511, - "pct_cuda_time": 0.6771420000364278, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 507.621, - "cuda_time_us": 2353.8179999999998, - "pct_cuda_time": 3.262708574099138, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.596, - "cuda_time_us": 33.824, - "pct_cuda_time": 0.04688461674196104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.824, - "pct_cuda_time": 0.04688461674196104, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 360.25, - "cuda_time_us": 574.655, - "pct_cuda_time": 0.7965491790992083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.028, - "cuda_time_us": 234.303, - "pct_cuda_time": 0.32477549540242723, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 233.567, - "pct_cuda_time": 0.3237553003361405, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.694, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.0611229913627458, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.0611229913627458, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.107, - "cuda_time_us": 129.152, - "pct_cuda_time": 0.17902205597971121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.024218543747503055, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 110.24, - "pct_cuda_time": 0.1528074784068645, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0019960338253436584, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.021, - "cuda_time_us": 167.10399999999998, - "pct_cuda_time": 0.23162863635432404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 166.368, - "pct_cuda_time": 0.2306084412880373, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.162, - "cuda_time_us": 32.608, - "pct_cuda_time": 0.045199077067226395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.608, - "pct_cuda_time": 0.045199077067226395, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.16, - "cuda_time_us": 1712.731, - "pct_cuda_time": 2.3740757011907423, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.311, - "cuda_time_us": 1085.788, - "pct_cuda_time": 1.5050483160779444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.407, - "pct_cuda_time": 0.0019502913835128663, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1084.381, - "pct_cuda_time": 1.5030980246944319, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.799, - "cuda_time_us": 139.136, - "pct_cuda_time": 0.19286122383542725, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 139.136, - "pct_cuda_time": 0.19286122383542725, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.08, - "cuda_time_us": 487.807, - "pct_cuda_time": 0.6761661612773708, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 487.071, - "pct_cuda_time": 0.675145966211084, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 492.288, - "cuda_time_us": 2360.217, - "pct_cuda_time": 3.2715784494105096, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.884, - "cuda_time_us": 34.08, - "pct_cuda_time": 0.04723946719979991, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.08, - "pct_cuda_time": 0.04723946719979991, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 351.291, - "cuda_time_us": 577.278, - "pct_cuda_time": 0.8001850101574559, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.641, - "cuda_time_us": 234.911, - "pct_cuda_time": 0.32561826523979454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.00190732121088394, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 233.535, - "pct_cuda_time": 0.32371094402891054, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 93.719, - "cuda_time_us": 43.872, - "pct_cuda_time": 0.060812497212136794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.872, - "pct_cuda_time": 0.060812497212136794, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.834, - "cuda_time_us": 130.016, - "pct_cuda_time": 0.1802196762749174, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 18.368, - "pct_cuda_time": 0.02546052034993911, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 110.208, - "pct_cuda_time": 0.15276312209963464, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0019960338253436584, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.287, - "cuda_time_us": 168.47899999999998, - "pct_cuda_time": 0.23353457143060707, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 167.391, - "pct_cuda_time": 0.23202645698479188, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.957, - "cuda_time_us": 33.344, - "pct_cuda_time": 0.04621927213351316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.344, - "pct_cuda_time": 0.04621927213351316, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.62, - "cuda_time_us": 1715.5149999999999, - "pct_cuda_time": 2.37793469991974, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.949, - "cuda_time_us": 1086.204, - "pct_cuda_time": 1.5056249480719326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0010188089316858257, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1085.469, - "pct_cuda_time": 1.504606139140247, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.258, - "cuda_time_us": 139.072, - "pct_cuda_time": 0.19277251122096756, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 139.072, - "pct_cuda_time": 0.19277251122096756, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.964, - "cuda_time_us": 490.239, - "pct_cuda_time": 0.6795372406268401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.152, - "pct_cuda_time": 0.0015968270602749264, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 489.087, - "pct_cuda_time": 0.6779404135665652, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 506.076, - "cuda_time_us": 2355.323, - "pct_cuda_time": 3.2647947066735425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.315, - "cuda_time_us": 34.08, - "pct_cuda_time": 0.04723946719979991, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.08, - "pct_cuda_time": 0.04723946719979991, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 363.145, - "cuda_time_us": 575.7429999999999, - "pct_cuda_time": 0.7980572935450234, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.16, - "cuda_time_us": 235.26399999999998, - "pct_cuda_time": 0.3261075707539239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 234.528, - "pct_cuda_time": 0.3250873756876371, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.883, - "cuda_time_us": 43.743, - "pct_cuda_time": 0.06063368584861643, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.743, - "pct_cuda_time": 0.06063368584861643, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.715, - "cuda_time_us": 129.248, - "pct_cuda_time": 0.1791551249014008, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.568, - "pct_cuda_time": 0.024351612669192634, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 110.048, - "pct_cuda_time": 0.15254134056348537, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.0022621716687228127, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.392, - "cuda_time_us": 167.488, - "pct_cuda_time": 0.23216091204108238, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 166.752, - "pct_cuda_time": 0.23114071697479566, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.589, - "cuda_time_us": 32.16, - "pct_cuda_time": 0.04457808876600836, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.16, - "pct_cuda_time": 0.04457808876600836, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.72, - "cuda_time_us": 1713.34, - "pct_cuda_time": 2.374919857162711, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.402, - "cuda_time_us": 1083.614, - "pct_cuda_time": 1.502034859455516, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1082.558, - "pct_cuda_time": 1.5005711013169305, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.797, - "cuda_time_us": 139.359, - "pct_cuda_time": 0.19317033185143534, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 139.359, - "pct_cuda_time": 0.19317033185143534, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.981, - "cuda_time_us": 490.367, - "pct_cuda_time": 0.6797146658557596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.0018629649036540812, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 489.023, - "pct_cuda_time": 0.6778517009521055, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 485.629, - "cuda_time_us": 2357.308, - "pct_cuda_time": 3.267546183856395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.887, - "cuda_time_us": 33.919, - "pct_cuda_time": 0.047016299529049684, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.919, - "pct_cuda_time": 0.047016299529049684, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 345.112, - "cuda_time_us": 576.672, - "pct_cuda_time": 0.7993450125892905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.374, - "cuda_time_us": 235.488, - "pct_cuda_time": 0.32641806490453296, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.0013750455241256314, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 234.496, - "pct_cuda_time": 0.3250430193804073, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.655, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.0605907156759875, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.0605907156759875, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 147.449, - "cuda_time_us": 129.985, - "pct_cuda_time": 0.18017670610228853, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.792, - "pct_cuda_time": 0.02466210681980165, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 110.592, - "pct_cuda_time": 0.15329539778639298, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.601, - "pct_cuda_time": 0.0022192014960938865, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.227, - "cuda_time_us": 167.48700000000002, - "pct_cuda_time": 0.23215952590648148, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0010188089316858257, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 166.752, - "pct_cuda_time": 0.23114071697479566, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.678, - "cuda_time_us": 32.928, - "pct_cuda_time": 0.04564264013952499, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.928, - "pct_cuda_time": 0.04564264013952499, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.371, - "cuda_time_us": 1713.789, - "pct_cuda_time": 2.37554223159853, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.809, - "cuda_time_us": 1083.711, - "pct_cuda_time": 1.5021693145118065, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0010215812008876918, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1082.974, - "pct_cuda_time": 1.5011477333109187, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.612, - "cuda_time_us": 140.255, - "pct_cuda_time": 0.1944123084538714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 140.255, - "pct_cuda_time": 0.1944123084538714, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.253, - "cuda_time_us": 489.82300000000004, - "pct_cuda_time": 0.678960608632852, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0015081144458152086, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 488.735, - "pct_cuda_time": 0.6774524941870367, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 483.816, - "cuda_time_us": 2341.7560000000003, - "pct_cuda_time": 3.245989018542684, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.82, - "cuda_time_us": 33.92, - "pct_cuda_time": 0.04701768566365062, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.92, - "pct_cuda_time": 0.04701768566365062, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 345.943, - "cuda_time_us": 575.583, - "pct_cuda_time": 0.7978355120088743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.181, - "cuda_time_us": 234.624, - "pct_cuda_time": 0.3252204446093267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 233.888, - "pct_cuda_time": 0.32420024954304, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 93.489, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.06072378459767707, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.06072378459767707, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.255, - "cuda_time_us": 129.24699999999999, - "pct_cuda_time": 0.17915373876679985, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.375, - "pct_cuda_time": 0.024084088691212546, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 110.463, - "pct_cuda_time": 0.15311658642287257, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.409, - "pct_cuda_time": 0.0019530636527147324, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.592, - "cuda_time_us": 167.90400000000002, - "pct_cuda_time": 0.2327375440350706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0014637581385853495, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 166.848, - "pct_cuda_time": 0.23127378589648523, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.477, - "cuda_time_us": 32.8, - "pct_cuda_time": 0.045465214910605546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.8, - "pct_cuda_time": 0.045465214910605546, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.482, - "cuda_time_us": 1699.4530000000002, - "pct_cuda_time": 2.3556706059595536, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.088, - "cuda_time_us": 1081.6940000000002, - "pct_cuda_time": 1.4993734810217247, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1080.958, - "pct_cuda_time": 1.4983532859554378, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.258, - "cuda_time_us": 138.24, - "pct_cuda_time": 0.1916192472329912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 138.24, - "pct_cuda_time": 0.1916192472329912, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.667, - "cuda_time_us": 479.51899999999995, - "pct_cuda_time": 0.6646778777048372, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 478.751, - "pct_cuda_time": 0.6636133263313206, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 492.363, - "cuda_time_us": 2321.9480000000003, - "pct_cuda_time": 3.218532464367401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.642, - "cuda_time_us": 34.464, - "pct_cuda_time": 0.04777174288655822, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.464, - "pct_cuda_time": 0.04777174288655822, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 349.559, - "cuda_time_us": 567.071, - "pct_cuda_time": 0.7860367342857317, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.67, - "cuda_time_us": 231.45600000000002, - "pct_cuda_time": 0.3208291701935707, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.00190732121088394, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 230.08, - "pct_cuda_time": 0.3189218489826868, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.824, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.05983665845307989, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.05983665845307989, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 149.646, - "cuda_time_us": 127.488, - "pct_cuda_time": 0.17671552800375853, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.664, - "pct_cuda_time": 0.024484681590882212, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 108.416, - "pct_cuda_time": 0.15027916889476253, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.0019516775181137992, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.508, - "cuda_time_us": 164.959, - "pct_cuda_time": 0.2286553776353226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 164.223, - "pct_cuda_time": 0.22763518256903584, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.213, - "cuda_time_us": 33.024, - "pct_cuda_time": 0.045775709061214566, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.024, - "pct_cuda_time": 0.045775709061214566, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.157, - "cuda_time_us": 1687.3890000000001, - "pct_cuda_time": 2.338948278133896, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.234, - "cuda_time_us": 1068.67, - "pct_cuda_time": 1.481320463979172, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1067.934, - "pct_cuda_time": 1.480300268912885, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.392, - "cuda_time_us": 138.528, - "pct_cuda_time": 0.19201845399805992, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 138.528, - "pct_cuda_time": 0.19201845399805992, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.364, - "cuda_time_us": 480.191, - "pct_cuda_time": 0.6656093601566643, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 479.167, - "pct_cuda_time": 0.6641899583253088, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 483.139, - "cuda_time_us": 2314.809, - "pct_cuda_time": 3.2086368494513393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.874, - "cuda_time_us": 34.048, - "pct_cuda_time": 0.04719511089257006, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.048, - "pct_cuda_time": 0.04719511089257006, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 343.571, - "cuda_time_us": 566.108, - "pct_cuda_time": 0.784701886665033, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.301, - "cuda_time_us": 230.68699999999998, - "pct_cuda_time": 0.3197632326854531, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 229.951, - "pct_cuda_time": 0.31874303761916634, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[3072, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 94.411, - "cuda_time_us": 43.743, - "pct_cuda_time": 0.06063368584861643, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.743, - "pct_cuda_time": 0.06063368584861643, - "trace": "_C::rotary_embedding(int64[3072], bfloat16[3072, 4096], bfloat16[3072, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 150.246, - "cuda_time_us": 127.135, - "pct_cuda_time": 0.1762262224896292, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 17.28, - "pct_cuda_time": 0.0239524059041239, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[3072], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 108.223, - "pct_cuda_time": 0.15001164491678246, - "trace": "_vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.0022621716687228127, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], None, None, bfloat16[3072, 32, 128], int32[3], int32[3], None, None, None, 1536, 1536, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[3072, 32, 128], bfloat16[3072, 8, 128], bfloat16[3072, 8, 128], bfloat16[3072, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.884, - "cuda_time_us": 164.543, - "pct_cuda_time": 0.22807874564133446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.375, - "pct_cuda_time": 0.0019059350762830071, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 163.168, - "pct_cuda_time": 0.22617281056505142, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[3072, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.786, - "cuda_time_us": 32.128, - "pct_cuda_time": 0.04453373245877851, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.128, - "pct_cuda_time": 0.04453373245877851, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.783, - "cuda_time_us": 1682.525, - "pct_cuda_time": 2.3322061194349577, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.554, - "cuda_time_us": 1063.678, - "pct_cuda_time": 1.474400880051314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.001064551373516618, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1062.91, - "pct_cuda_time": 1.4733363286777972, - "trace": "mm(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[3072, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[3072, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.519, - "cuda_time_us": 138.208, - "pct_cuda_time": 0.19157489092576133, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 138.208, - "pct_cuda_time": 0.19157489092576133, - "trace": "_C::silu_and_mul(bfloat16[3072, 14336], bfloat16[3072, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.823, - "cuda_time_us": 480.639, - "pct_cuda_time": 0.6662303484578824, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0010201950662867589, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 479.903, - "pct_cuda_time": 0.6652101533915956, - "trace": "mm(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[3072, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[3072, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.661, - "cuda_time_us": 34.368, - "pct_cuda_time": 0.04763867396486865, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.368, - "pct_cuda_time": 0.04763867396486865, - "trace": "_C::fused_add_rms_norm(bfloat16[3072, 4096], bfloat16[3072, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 91.039, - "cuda_time_us": 358.335, - "pct_cuda_time": 0.496700542225361, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.005145331638663653, - "trace": "index_select(bfloat16[3072, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0014194018313554906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 353.599, - "pct_cuda_time": 0.49013580875534185, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 56452.16, - "cuda_time_us": 139.584, - "pct_cuda_time": 0.1934822121366453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.0029718725844005583, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.0029718725844005583, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.003016228891630417, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.003016228891630417, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.0029718725844005583, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.003016228891630417, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.003016228891630417, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.672, - "pct_cuda_time": 0.006476020855559424, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 5.312, - "pct_cuda_time": 0.007363147000156607, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 39.36, - "pct_cuda_time": 0.054558257892726655, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 32.576, - "pct_cuda_time": 0.04515472075999654, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0031049415060901355, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 5.216, - "pct_cuda_time": 0.00723007807846703, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 32.768, - "pct_cuda_time": 0.0454208586033757, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.003193654120549853, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - }, - "decode_1": { - "metadata": { - "num_running_seqs": 2 - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 6667.974, - "pct_cuda_time": 93.3759268136013, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 6661.126, - "pct_cuda_time": 93.2800298669696, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 211.6480000000001, - "pct_cuda_time": 2.963843014121695, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.384, - "pct_cuda_time": 0.06139197050720775, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 207.26400000000012, - "pct_cuda_time": 2.9024510436144877, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 2198.167, - "pct_cuda_time": 30.78234571941545, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 671.329, - "pct_cuda_time": 9.401051589560508, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 671.329, - "pct_cuda_time": 9.401051589560508, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 119.07000000000002, - "pct_cuda_time": 1.6674137610157909, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 119.07000000000002, - "pct_cuda_time": 1.6674137610157909, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 836.5099999999998, - "pct_cuda_time": 11.714187328691683, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 73.72800000000001, - "pct_cuda_time": 1.0324605842963992, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cuda_time_us": 638.3969999999999, - "pct_cuda_time": 8.93988362132525, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cuda_time_us": 78.657, - "pct_cuda_time": 1.1014845401882845, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 45.72800000000001, - "pct_cuda_time": 0.640358582881751, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 571.2579999999999, - "pct_cuda_time": 7.999693040147463, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cuda_time_us": 503.48399999999987, - "pct_cuda_time": 7.050610145723307, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cuda_time_us": 67.77400000000002, - "pct_cuda_time": 0.9490828944241558, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 4251.310999999999, - "pct_cuda_time": 59.53384113343244, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 2577.2090000000003, - "pct_cuda_time": 36.09031453442299, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2577.2090000000003, - "pct_cuda_time": 36.09031453442299, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 283.87, - "pct_cuda_time": 3.9752141121991476, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 283.87, - "pct_cuda_time": 3.9752141121991476, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 1390.232, - "pct_cuda_time": 19.46831248681032, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1390.232, - "pct_cuda_time": 19.46831248681032, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0483965898888937, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0483965898888937, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 349.279, - "pct_cuda_time": 4.891178391146673, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.072, - "pct_cuda_time": 0.04301919101234995, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.768, - "pct_cuda_time": 0.010754797753087488, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 345.439, - "pct_cuda_time": 4.837404402381236, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 123.746, - "pct_cuda_time": 1.7328947952520368, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 15.424000000000001, - "pct_cuda_time": 0.2159921882078404, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.16, - "pct_cuda_time": 0.05825515449589056, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.577, - "pct_cuda_time": 0.06409467358838727, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 34.271, - "pct_cuda_time": 0.4799188460886215, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.936, - "pct_cuda_time": 0.3912057682685574, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 1.825, - "pct_cuda_time": 0.025556648306490452, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.833, - "pct_cuda_time": 0.06767960617274978, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 28.256, - "pct_cuda_time": 0.39568693399901056, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 17573.348, - "cuda_time_us": 6667.974, - "pct_cuda_time": 93.3759268136013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 70.593, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[2]) <- embedding(bfloat16[128256, 4096], int64[2], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 981.958, - "cuda_time_us": 211.551, - "pct_cuda_time": 2.96248466075965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 64.841, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.06139197050720775, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.384, - "pct_cuda_time": 0.06139197050720775, - "trace": "_C::rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 710.481, - "cuda_time_us": 71.10399999999998, - "pct_cuda_time": 0.9957150253066831, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 94.622, - "cuda_time_us": 23.391, - "pct_cuda_time": 0.3275592112532154, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 23.391, - "pct_cuda_time": 0.3275592112532154, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 223.417, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05018905618107494, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05018905618107494, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 270.375, - "cuda_time_us": 26.369, - "pct_cuda_time": 0.3692620598322448, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.497, - "pct_cuda_time": 0.034967096340442004, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.744, - "pct_cuda_time": 0.2764879255689575, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.023302061798356224, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.147, - "cuda_time_us": 17.759999999999998, - "pct_cuda_time": 0.24870469804014814, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.21868088764611227, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.030023810394035906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.757, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 149.842, - "cuda_time_us": 132.927, - "pct_cuda_time": 1.8614622407873185, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 52.717, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1216357823324161, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1216357823324161, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 40.718, - "cuda_time_us": 9.472, - "pct_cuda_time": 0.13264250562141233, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.472, - "pct_cuda_time": 0.13264250562141233, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 36.183, - "cuda_time_us": 43.359, - "pct_cuda_time": 0.6071839528334901, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.359, - "pct_cuda_time": 0.6071839528334901, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 579.584, - "cuda_time_us": 210.43300000000002, - "pct_cuda_time": 2.94682858798888, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.601, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 416.647, - "cuda_time_us": 69.47300000000001, - "pct_cuda_time": 0.9728750837242802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.687, - "cuda_time_us": 21.632, - "pct_cuda_time": 0.302926803378631, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.632, - "pct_cuda_time": 0.302926803378631, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 118.835, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 190.274, - "cuda_time_us": 26.337, - "pct_cuda_time": 0.3688139432591994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.273, - "pct_cuda_time": 0.03183028032912482, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.16, - "pct_cuda_time": 0.28231344101854655, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.0336087429783984, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.843, - "cuda_time_us": 17.728, - "pct_cuda_time": 0.2482565814671029, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.21868088764611227, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.463, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 107.491, - "cuda_time_us": 134.656, - "pct_cuda_time": 1.885674539374673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.352, - "cuda_time_us": 81.92, - "pct_cuda_time": 1.1471784269959988, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.92, - "pct_cuda_time": 1.1471784269959988, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 23.272, - "cuda_time_us": 9.152, - "pct_cuda_time": 0.12816133989095924, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.152, - "pct_cuda_time": 0.12816133989095924, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.381, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.610334772487715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.610334772487715, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 510.899, - "cuda_time_us": 208.22200000000004, - "pct_cuda_time": 2.9158665335200307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.567, - "cuda_time_us": 3.295, - "pct_cuda_time": 0.04614200338075947, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.295, - "pct_cuda_time": 0.04614200338075947, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 360.698, - "cuda_time_us": 68.095, - "pct_cuda_time": 0.9535780637975163, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.386, - "cuda_time_us": 20.703, - "pct_cuda_time": 0.2899174191174092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.703, - "pct_cuda_time": 0.2899174191174092, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.297, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 163.665, - "cuda_time_us": 26.176, - "pct_cuda_time": 0.3665593567510652, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.034953092697534334, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.84, - "pct_cuda_time": 0.27783227528809346, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.034953092697534334, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.018820896067903103, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.405, - "cuda_time_us": 17.439999999999998, - "pct_cuda_time": 0.24422353230969504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.21464783848870445, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.734, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.307, - "cuda_time_us": 133.60000000000002, - "pct_cuda_time": 1.870886692464178, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.88, - "cuda_time_us": 81.312, - "pct_cuda_time": 1.1386642121081378, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.312, - "pct_cuda_time": 1.1386642121081378, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.509, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12368017416050613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12368017416050613, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.174, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.6085423061955338, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.6085423061955338, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 519.986, - "cuda_time_us": 207.998, - "pct_cuda_time": 2.9127297175087126, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.816, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04660412359671245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04660412359671245, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 376.259, - "cuda_time_us": 68.31899999999999, - "pct_cuda_time": 0.9567148798088333, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.685, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.2903795393333622, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.2903795393333622, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 105.097, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.050637172754120253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.050637172754120253, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 172.28, - "cuda_time_us": 25.951, - "pct_cuda_time": 0.3634085370968404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.303, - "pct_cuda_time": 0.0322503896163548, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.904, - "pct_cuda_time": 0.27872850843418406, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03405685955144371, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018372779494857792, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.456, - "cuda_time_us": 18.016, - "pct_cuda_time": 0.25228963062451065, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.904, - "pct_cuda_time": 0.22271393680352009, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.454, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.541, - "cuda_time_us": 133.119, - "pct_cuda_time": 1.8641509402255902, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.519, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1279094143550503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1279094143550503, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.416, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12368017416050613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12368017416050613, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.863, - "cuda_time_us": 43.743, - "pct_cuda_time": 0.6125613517100339, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.743, - "pct_cuda_time": 0.6125613517100339, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 491.022, - "cuda_time_us": 209.82399999999998, - "pct_cuda_time": 2.9383003694581107, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.955, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04705224016975776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04705224016975776, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 350.997, - "cuda_time_us": 68.8, - "pct_cuda_time": 0.9634506320474208, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.712, - "cuda_time_us": 21.12, - "pct_cuda_time": 0.29575693820990595, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.12, - "pct_cuda_time": 0.29575693820990595, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.804, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.33, - "cuda_time_us": 26.144, - "pct_cuda_time": 0.36611124017801994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.031368160113171846, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.904, - "pct_cuda_time": 0.27872850843418406, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.034953092697534334, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.404, - "cuda_time_us": 17.759999999999998, - "pct_cuda_time": 0.24870469804014814, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.648, - "pct_cuda_time": 0.21912900421915757, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.572, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.045273777520484185, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.045273777520484185, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.393, - "cuda_time_us": 134.43099999999998, - "pct_cuda_time": 1.882523719720448, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.106, - "cuda_time_us": 81.823, - "pct_cuda_time": 1.145820073633955, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.823, - "pct_cuda_time": 1.145820073633955, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.197, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.12143959129527956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.12143959129527956, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.254, - "cuda_time_us": 43.936, - "pct_cuda_time": 0.6152640547912134, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.936, - "pct_cuda_time": 0.6152640547912134, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 548.317, - "cuda_time_us": 207.68200000000002, - "pct_cuda_time": 2.9083045663498908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.73, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.04974093960802964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.04974093960802964, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 395.241, - "cuda_time_us": 68.19200000000001, - "pct_cuda_time": 0.95493641715956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.993, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.29127577247945285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.29127577247945285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.023, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05198152247325619, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05198152247325619, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 171.134, - "cuda_time_us": 26.208, - "pct_cuda_time": 0.3670074733241105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03226439325926246, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.872, - "pct_cuda_time": 0.2782803918611388, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.03540120927057965, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.286, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.24467164888274037, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.21464783848870445, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.030023810394035906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.858, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04572189409352949, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04572189409352949, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.059, - "cuda_time_us": 132.673, - "pct_cuda_time": 1.8579053154887712, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 38.465, - "cuda_time_us": 80.737, - "pct_cuda_time": 1.1306121174362298, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.737, - "pct_cuda_time": 1.1306121174362298, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.582, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12323205758746081, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12323205758746081, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.743, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.6040611404650806, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.6040611404650806, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 504.018, - "cuda_time_us": 207.519, - "pct_cuda_time": 2.906021972555941, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.035, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.04794847331584839, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.04794847331584839, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 363.796, - "cuda_time_us": 68.607, - "pct_cuda_time": 0.9607479289662413, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.412, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.29127577247945285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.29127577247945285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.423, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05018905618107494, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05018905618107494, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.661, - "cuda_time_us": 25.790999999999997, - "pct_cuda_time": 0.3611679542316138, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03226439325926246, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.711, - "pct_cuda_time": 0.2760258053530045, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018372779494857792, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.248, - "cuda_time_us": 18.432000000000002, - "pct_cuda_time": 0.2581151460740998, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.352, - "pct_cuda_time": 0.22898756882615443, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.02912757724794528, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.298, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.727, - "cuda_time_us": 132.32, - "pct_cuda_time": 1.852962029542365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.955, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1279094143550503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1279094143550503, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.649, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12502452387964208, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12502452387964208, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.489, - "cuda_time_us": 42.848, - "pct_cuda_time": 0.6000280913076728, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.848, - "pct_cuda_time": 0.6000280913076728, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 487.779, - "cuda_time_us": 207.551, - "pct_cuda_time": 2.906470089128986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.292, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05018905618107494, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05018905618107494, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 349.351, - "cuda_time_us": 68.32, - "pct_cuda_time": 0.956728883451741, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.831, - "cuda_time_us": 21.153, - "pct_cuda_time": 0.2962190584258589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.153, - "pct_cuda_time": 0.2962190584258589, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 105.752, - "cuda_time_us": 3.711, - "pct_cuda_time": 0.05196751883034852, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.711, - "pct_cuda_time": 0.05196751883034852, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.632, - "cuda_time_us": 25.855999999999998, - "pct_cuda_time": 0.3620781910206121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.872, - "pct_cuda_time": 0.2782803918611388, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03405685955144371, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.01792466292181248, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.848, - "cuda_time_us": 17.6, - "pct_cuda_time": 0.24646411517492162, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.488, - "pct_cuda_time": 0.21688842135393102, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.357, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.9, - "cuda_time_us": 132.351, - "pct_cuda_time": 1.8533961424725027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.242, - "cuda_time_us": 80.159, - "pct_cuda_time": 1.1225180118355989, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.159, - "pct_cuda_time": 1.1225180118355989, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 19.725, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.12054335814918894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.12054335814918894, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.686, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.610334772487715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.610334772487715, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 502.588, - "cuda_time_us": 209.729, - "pct_cuda_time": 2.9369700233818827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.752, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 357.922, - "cuda_time_us": 70.145, - "pct_cuda_time": 0.9822855317582315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.893, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.30113433708644965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.30113433708644965, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.796, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.053325872192392126, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.053325872192392126, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.948, - "cuda_time_us": 26.240999999999996, - "pct_cuda_time": 0.3674695935400635, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.031368160113171846, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.935, - "pct_cuda_time": 0.2791626213643217, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.561, - "pct_cuda_time": 0.03586332948653263, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.02107548257603733, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.752, - "cuda_time_us": 18.592, - "pct_cuda_time": 0.2603557289393263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.48, - "pct_cuda_time": 0.23078003511833572, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.38, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.847, - "cuda_time_us": 133.12, - "pct_cuda_time": 1.864164943868498, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.946, - "cuda_time_us": 81.504, - "pct_cuda_time": 1.1413529115464096, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.504, - "pct_cuda_time": 1.1413529115464096, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.3, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.12009524157614362, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.12009524157614362, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.362, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.6027167907459446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.6027167907459446, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 545.392, - "cuda_time_us": 207.261, - "pct_cuda_time": 2.902409032685763, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.896, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048844706461939, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048844706461939, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 393.751, - "cuda_time_us": 68.83, - "pct_cuda_time": 0.9638707413346508, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.151, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.2894833061872716, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.2894833061872716, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.305, - "cuda_time_us": 3.775, - "pct_cuda_time": 0.05286375197643916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.775, - "pct_cuda_time": 0.05286375197643916, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 165.89, - "cuda_time_us": 26.623, - "pct_cuda_time": 0.3728189851307919, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.303, - "pct_cuda_time": 0.0322503896163548, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.352, - "pct_cuda_time": 0.28500214045681843, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.868, - "cuda_time_us": 17.759999999999998, - "pct_cuda_time": 0.24870469804014814, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.488, - "pct_cuda_time": 0.21688842135393102, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.92, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.203, - "cuda_time_us": 131.647, - "pct_cuda_time": 1.8435375778655057, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.86, - "cuda_time_us": 79.807, - "pct_cuda_time": 1.1175887295321005, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.807, - "pct_cuda_time": 1.1175887295321005, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.862, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.12143959129527956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.12143959129527956, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.248, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6045092570381259, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6045092570381259, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 495.799, - "cuda_time_us": 207.103, - "pct_cuda_time": 2.9001964571063517, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.47, - "cuda_time_us": 3.425, - "pct_cuda_time": 0.04796247695875605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.425, - "pct_cuda_time": 0.04796247695875605, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 350.927, - "cuda_time_us": 68.12700000000001, - "pct_cuda_time": 0.9540261803705617, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.143, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.29306823877163407, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.29306823877163407, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.814, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.04974093960802964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.04974093960802964, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.083, - "cuda_time_us": 26.015, - "pct_cuda_time": 0.364304770242931, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.031368160113171846, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.032, - "pct_cuda_time": 0.28052097472636534, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.431, - "pct_cuda_time": 0.03404285590853605, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018372779494857792, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.829, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.24691223174796692, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.21733653792697633, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.33, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.043901420515532905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.043901420515532905, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.827, - "cuda_time_us": 132.416, - "pct_cuda_time": 1.854306379261501, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.795, - "cuda_time_us": 79.776, - "pct_cuda_time": 1.1171546166019628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.776, - "pct_cuda_time": 1.1171546166019628, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.653, - "cuda_time_us": 9.28, - "pct_cuda_time": 0.12995380618314048, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.28, - "pct_cuda_time": 0.12995380618314048, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.621, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.6071979564763977, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.6071979564763977, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 537.959, - "cuda_time_us": 208.191, - "pct_cuda_time": 2.9154324205898927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.392, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 391.732, - "cuda_time_us": 68.864, - "pct_cuda_time": 0.9643468651935115, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.424, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.28724272332204503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.28724272332204503, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.653, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05198152247325619, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05198152247325619, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 165.521, - "cuda_time_us": 26.304, - "pct_cuda_time": 0.36835182304324643, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.064, - "pct_cuda_time": 0.2809690912994106, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018372779494857792, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 36.994, - "cuda_time_us": 18.336, - "pct_cuda_time": 0.2567707963549638, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.256, - "pct_cuda_time": 0.22764321910701849, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.02912757724794528, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.411, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.417, - "cuda_time_us": 132.767, - "pct_cuda_time": 1.859221657922092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.587, - "cuda_time_us": 80.319, - "pct_cuda_time": 1.1247585947008256, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.319, - "pct_cuda_time": 1.1247585947008256, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.765, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12412829073355144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12412829073355144, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.566, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.610334772487715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.610334772487715, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 558.98, - "cuda_time_us": 207.327, - "pct_cuda_time": 2.903333273117669, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.146, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 403.864, - "cuda_time_us": 68.673, - "pct_cuda_time": 0.9616721693981473, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.468, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.28724272332204503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.28724272332204503, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.249, - "cuda_time_us": 3.713, - "pct_cuda_time": 0.051995526116163863, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.713, - "pct_cuda_time": 0.051995526116163863, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 206.059, - "cuda_time_us": 26.784000000000002, - "pct_cuda_time": 0.3750735716389262, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.031368160113171846, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.28545025702986376, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.696, - "pct_cuda_time": 0.023750178371401535, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 37.658, - "cuda_time_us": 17.664, - "pct_cuda_time": 0.24736034832101225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.21733653792697633, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.030023810394035906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.822, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 107.291, - "cuda_time_us": 132.126, - "pct_cuda_time": 1.850245322818278, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.12, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1279094143550503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1279094143550503, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.945, - "cuda_time_us": 8.959, - "pct_cuda_time": 0.12545863680977967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.959, - "pct_cuda_time": 0.12545863680977967, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 36.625, - "cuda_time_us": 42.623, - "pct_cuda_time": 0.596877271653448, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.623, - "pct_cuda_time": 0.596877271653448, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 499.868, - "cuda_time_us": 208.673, - "pct_cuda_time": 2.9221821764713876, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.975, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 360.272, - "cuda_time_us": 68.64, - "pct_cuda_time": 0.9612100491821942, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.425, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.2899314227603169, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.2899314227603169, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 108.991, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.45, - "cuda_time_us": 26.274, - "pct_cuda_time": 0.3679317137560165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.033, - "pct_cuda_time": 0.280534978369273, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.465, - "pct_cuda_time": 0.03451897976739669, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.823, - "cuda_time_us": 17.886, - "pct_cuda_time": 0.25046915704651407, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.743, - "pct_cuda_time": 0.22045935029538585, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.030009806751128236, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.171, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04572189409352949, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04572189409352949, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.168, - "cuda_time_us": 133.47199999999998, - "pct_cuda_time": 1.869094226171996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.815, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1337349298046393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1337349298046393, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.669, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.1268169901718233, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.1268169901718233, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.989, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.6085423061955338, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.6085423061955338, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 506.802, - "cuda_time_us": 207.51799999999997, - "pct_cuda_time": 2.9060079689130327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.923, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 350.848, - "cuda_time_us": 68.223, - "pct_cuda_time": 0.9553705300896975, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.622, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.2903795393333622, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.2903795393333622, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.247, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05108528932716558, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05108528932716558, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.162, - "cuda_time_us": 25.919, - "pct_cuda_time": 0.36296042052379507, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.271, - "pct_cuda_time": 0.03180227304330949, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.904, - "pct_cuda_time": 0.27872850843418406, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.0336087429783984, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.018820896067903103, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.974, - "cuda_time_us": 17.92, - "pct_cuda_time": 0.25094528090537477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.84, - "pct_cuda_time": 0.22181770365742945, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.02912757724794528, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.809, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 109.388, - "cuda_time_us": 132.767, - "pct_cuda_time": 1.859221657922092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.961, - "cuda_time_us": 80.799, - "pct_cuda_time": 1.1314803432965053, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.799, - "pct_cuda_time": 1.1314803432965053, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.016, - "cuda_time_us": 8.609, - "pct_cuda_time": 0.12055736179209661, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.609, - "pct_cuda_time": 0.12055736179209661, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.665, - "cuda_time_us": 43.359, - "pct_cuda_time": 0.6071839528334901, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.359, - "pct_cuda_time": 0.6071839528334901, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 491.246, - "cuda_time_us": 207.17, - "pct_cuda_time": 2.901134701181165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.313, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.04751436038571073, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.393, - "pct_cuda_time": 0.04751436038571073, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 351.117, - "cuda_time_us": 68.417, - "pct_cuda_time": 0.9580872368137847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.563, - "cuda_time_us": 21.152, - "pct_cuda_time": 0.2962050547829513, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.152, - "pct_cuda_time": 0.2962050547829513, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.223, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.052429639046301504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.052429639046301504, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.931, - "cuda_time_us": 25.855999999999998, - "pct_cuda_time": 0.3620781910206121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03271250983230778, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.776, - "pct_cuda_time": 0.27693604214200285, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03405685955144371, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018372779494857792, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.259, - "cuda_time_us": 17.665, - "pct_cuda_time": 0.24737435196391988, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.552, - "pct_cuda_time": 0.21778465450002163, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.02958969746389826, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.177, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.694, - "cuda_time_us": 132.224, - "pct_cuda_time": 1.8516176798232291, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.119, - "cuda_time_us": 80.128, - "pct_cuda_time": 1.1220838989054613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.128, - "pct_cuda_time": 1.1220838989054613, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.224, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.12143959129527956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.12143959129527956, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.732, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.6080941896224884, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.6080941896224884, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 503.285, - "cuda_time_us": 209.18200000000002, - "pct_cuda_time": 2.9293100307113895, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.259, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 351.445, - "cuda_time_us": 70.046, - "pct_cuda_time": 0.9808991711103727, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.422, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.30113433708644965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.504, - "pct_cuda_time": 0.30113433708644965, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.422, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.054222105338482755, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.054222105338482755, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.208, - "cuda_time_us": 26.238, - "pct_cuda_time": 0.36742758261134056, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.367, - "pct_cuda_time": 0.03314662276244542, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.936, - "pct_cuda_time": 0.2791766250072294, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03405685955144371, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.021047475290222, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.108, - "cuda_time_us": 18.432000000000002, - "pct_cuda_time": 0.2581151460740998, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.32, - "pct_cuda_time": 0.22853945225310915, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.573, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.761, - "cuda_time_us": 132.736, - "pct_cuda_time": 1.8587875449919542, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.592, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1337349298046393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1337349298046393, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 27.888, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.1259207570257327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.1259207570257327, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.983, - "cuda_time_us": 42.784, - "pct_cuda_time": 0.5991318581615821, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.784, - "pct_cuda_time": 0.5991318581615821, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 501.532, - "cuda_time_us": 209.345, - "pct_cuda_time": 2.931592624505339, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.593, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 355.574, - "cuda_time_us": 69.47299999999998, - "pct_cuda_time": 0.9728750837242799, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.8, - "cuda_time_us": 21.697, - "pct_cuda_time": 0.3038370401676292, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.697, - "pct_cuda_time": 0.3038370401676292, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.617, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.055566455057618695, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.055566455057618695, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.081, - "cuda_time_us": 26.176, - "pct_cuda_time": 0.3665593567510652, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03226439325926246, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.904, - "pct_cuda_time": 0.27872850843418406, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.812, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.24691223174796692, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.488, - "pct_cuda_time": 0.21688842135393102, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.030023810394035906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.804, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.148, - "cuda_time_us": 133.28, - "pct_cuda_time": 1.8664055267337245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.96, - "cuda_time_us": 80.512, - "pct_cuda_time": 1.127461297782005, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.512, - "pct_cuda_time": 1.127461297782005, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.478, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12412829073355144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12412829073355144, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.242, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.6148159382181682, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.6148159382181682, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 527.232, - "cuda_time_us": 207.04000000000002, - "pct_cuda_time": 2.899314227603169, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.518, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.04794847331584839, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.04794847331584839, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 377.083, - "cuda_time_us": 68.224, - "pct_cuda_time": 0.9553845337326052, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.017, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.29217200562554346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.29217200562554346, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 114.722, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 167.776, - "cuda_time_us": 25.887999999999998, - "pct_cuda_time": 0.3625263075936574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.031368160113171846, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.904, - "pct_cuda_time": 0.27872850843418406, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.0336087429783984, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.018820896067903103, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.449, - "cuda_time_us": 17.792, - "pct_cuda_time": 0.2491528146131935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.68, - "pct_cuda_time": 0.2195771207922029, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.545, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.696, - "cuda_time_us": 132.224, - "pct_cuda_time": 1.8516176798232291, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.542, - "cuda_time_us": 79.84, - "pct_cuda_time": 1.1180508497480535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.84, - "pct_cuda_time": 1.1180508497480535, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.938, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12457640730659675, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12457640730659675, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.855, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.6089904227685791, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.6089904227685791, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 647.614, - "cuda_time_us": 206.88, - "pct_cuda_time": 2.8970736447379424, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.78, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04615600702366714, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 505.784, - "cuda_time_us": 68.257, - "pct_cuda_time": 0.9558466539485583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.474, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.28634649017595437, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.448, - "pct_cuda_time": 0.28634649017595437, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.237, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 312.03, - "cuda_time_us": 25.889, - "pct_cuda_time": 0.36254031123656505, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.273, - "pct_cuda_time": 0.03183028032912482, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.872, - "pct_cuda_time": 0.2782803918611388, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.0336087429783984, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.018820896067903103, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 37.78, - "cuda_time_us": 18.240000000000002, - "pct_cuda_time": 0.2554264466358279, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.128, - "pct_cuda_time": 0.22585075281483727, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.414, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04570789045062182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04570789045062182, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.235, - "cuda_time_us": 132.063, - "pct_cuda_time": 1.8493630933150949, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.684, - "cuda_time_us": 80.48, - "pct_cuda_time": 1.1270131812089599, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.48, - "pct_cuda_time": 1.1270131812089599, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.512, - "cuda_time_us": 8.607, - "pct_cuda_time": 0.12052935450628127, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.607, - "pct_cuda_time": 0.12052935450628127, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.729, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.601820557599854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.601820557599854, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 525.33, - "cuda_time_us": 208.79999999999998, - "pct_cuda_time": 2.9239606391206605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.173, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.053773988765437444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.053773988765437444, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 380.805, - "cuda_time_us": 68.32, - "pct_cuda_time": 0.956728883451741, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.392, - "cuda_time_us": 20.544, - "pct_cuda_time": 0.2876908398950903, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.544, - "pct_cuda_time": 0.2876908398950903, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.33, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.052877755619346815, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 176.533, - "cuda_time_us": 26.08, - "pct_cuda_time": 0.3652150070319293, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.808, - "pct_cuda_time": 0.2773841587150481, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.034953092697534334, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.161, - "cuda_time_us": 17.92, - "pct_cuda_time": 0.25094528090537477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.808, - "pct_cuda_time": 0.22136958708438415, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.582, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.506, - "cuda_time_us": 133.504, - "pct_cuda_time": 1.8695423427450415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.604, - "cuda_time_us": 80.704, - "pct_cuda_time": 1.1301499972202766, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.704, - "pct_cuda_time": 1.1301499972202766, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.232, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1227839410144155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1227839410144155, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.544, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.6166084045103493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.6166084045103493, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 498.994, - "cuda_time_us": 207.51999999999998, - "pct_cuda_time": 2.9060359761988486, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.038, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 358.696, - "cuda_time_us": 68.096, - "pct_cuda_time": 0.953592067440424, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.555, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.28903518961422625, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.28903518961422625, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.207, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05198152247325619, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05198152247325619, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 165.326, - "cuda_time_us": 26.304, - "pct_cuda_time": 0.36835182304324643, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03271250983230778, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.0, - "pct_cuda_time": 0.28007285815332, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.08, - "cuda_time_us": 17.439999999999998, - "pct_cuda_time": 0.24422353230969504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.21464783848870445, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.267, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.044811657304531204, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.044811657304531204, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.892, - "cuda_time_us": 132.832, - "pct_cuda_time": 1.8601318947110903, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.2, - "cuda_time_us": 79.968, - "pct_cuda_time": 1.1198433160402348, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.968, - "pct_cuda_time": 1.1198433160402348, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.766, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12233582444137019, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12233582444137019, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.453, - "cuda_time_us": 44.128, - "pct_cuda_time": 0.6179527542294853, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.128, - "pct_cuda_time": 0.6179527542294853, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 508.848, - "cuda_time_us": 208.317, - "pct_cuda_time": 2.917196879596258, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.17, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.047038236526850095, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.047038236526850095, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 360.604, - "cuda_time_us": 68.543, - "pct_cuda_time": 0.9598516958201507, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.426, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.29127577247945285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.29127577247945285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.865, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.054670221911528066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.054670221911528066, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.236, - "cuda_time_us": 26.24, - "pct_cuda_time": 0.3674555898971558, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.224, - "pct_cuda_time": 0.2832096741646372, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03405685955144371, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018372779494857792, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.69, - "cuda_time_us": 17.599, - "pct_cuda_time": 0.24645011153201393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.488, - "pct_cuda_time": 0.21688842135393102, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.029561690178082932, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.809, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.839, - "cuda_time_us": 133.279, - "pct_cuda_time": 1.8663915230908168, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.621, - "cuda_time_us": 80.287, - "pct_cuda_time": 1.1243104781277804, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.287, - "pct_cuda_time": 1.1243104781277804, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.232, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.13040192275618578, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.13040192275618578, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.317, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.611679122206851, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.611679122206851, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 511.776, - "cuda_time_us": 207.584, - "pct_cuda_time": 2.906932209344939, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.33, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 365.472, - "cuda_time_us": 67.96799999999999, - "pct_cuda_time": 0.9517996011482425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.848, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.2903795393333622, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.2903795393333622, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 105.003, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.873, - "cuda_time_us": 25.919999999999998, - "pct_cuda_time": 0.36297442416670267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03226439325926246, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.712, - "pct_cuda_time": 0.2760398089959122, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03405685955144371, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.02061336236008435, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.694, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.24691223174796692, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.21733653792697633, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.848, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.406, - "cuda_time_us": 133.312, - "pct_cuda_time": 1.86685364330677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.061, - "cuda_time_us": 80.768, - "pct_cuda_time": 1.1310462303663675, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.768, - "pct_cuda_time": 1.1310462303663675, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.29, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12368017416050613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12368017416050613, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.643, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6121272387798963, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6121272387798963, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 512.224, - "cuda_time_us": 207.932, - "pct_cuda_time": 2.9118054770768067, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.651, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.047038236526850095, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.047038236526850095, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 359.753, - "cuda_time_us": 68.79899999999999, - "pct_cuda_time": 0.9634366284045132, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.475, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.29486070506381534, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.29486070506381534, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.312, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.054222105338482755, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.054222105338482755, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.621, - "cuda_time_us": 26.08, - "pct_cuda_time": 0.3652150070319293, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.776, - "pct_cuda_time": 0.27693604214200285, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.03540120927057965, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.604, - "cuda_time_us": 17.791, - "pct_cuda_time": 0.2491388109702858, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.679, - "pct_cuda_time": 0.21956311714929522, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.146, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.769, - "cuda_time_us": 132.606, - "pct_cuda_time": 1.8569670714139577, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.309, - "cuda_time_us": 80.031, - "pct_cuda_time": 1.1207255455434177, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.031, - "pct_cuda_time": 1.1207255455434177, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.275, - "cuda_time_us": 9.216, - "pct_cuda_time": 0.12905757303704984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.216, - "pct_cuda_time": 0.12905757303704984, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.483, - "cuda_time_us": 43.359, - "pct_cuda_time": 0.6071839528334901, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.359, - "pct_cuda_time": 0.6071839528334901, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 508.298, - "cuda_time_us": 206.78300000000002, - "pct_cuda_time": 2.895715291375899, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.935, - "cuda_time_us": 3.295, - "pct_cuda_time": 0.04614200338075947, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.295, - "pct_cuda_time": 0.04614200338075947, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 362.664, - "cuda_time_us": 68.193, - "pct_cuda_time": 0.9549504208024676, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.941, - "cuda_time_us": 20.673, - "pct_cuda_time": 0.2894973098301792, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.673, - "pct_cuda_time": 0.2894973098301792, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.453, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.258, - "cuda_time_us": 26.112, - "pct_cuda_time": 0.36566312360497455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.872, - "pct_cuda_time": 0.2782803918611388, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.426, - "cuda_time_us": 17.728, - "pct_cuda_time": 0.2482565814671029, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.21868088764611227, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.743, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.443, - "cuda_time_us": 132.127, - "pct_cuda_time": 1.8502593264611857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.048, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1216357823324161, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1216357823324161, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.492, - "cuda_time_us": 9.024, - "pct_cuda_time": 0.126368873598778, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.024, - "pct_cuda_time": 0.126368873598778, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.58, - "cuda_time_us": 43.007, - "pct_cuda_time": 0.6022546705299917, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.007, - "pct_cuda_time": 0.6022546705299917, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 554.499, - "cuda_time_us": 208.702, - "pct_cuda_time": 2.9225882821157096, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.809, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04705224016975776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04705224016975776, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 410.578, - "cuda_time_us": 68.287, - "pct_cuda_time": 0.9562667632357883, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.426, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.28903518961422625, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.28903518961422625, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.675, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05108528932716558, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05108528932716558, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 204.026, - "cuda_time_us": 26.366999999999997, - "pct_cuda_time": 0.36923405254642944, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.368, - "pct_cuda_time": 0.03316062640535308, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.191, - "pct_cuda_time": 0.2827475539486842, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.018820896067903103, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.665, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.24691223174796692, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.21733653792697633, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.679, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.045259773877576515, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.538, - "cuda_time_us": 133.82299999999998, - "pct_cuda_time": 1.8740095048325869, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.916, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1337349298046393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1337349298046393, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.305, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12188770786832488, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12188770786832488, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.53, - "cuda_time_us": 44.159, - "pct_cuda_time": 0.618386867159623, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.159, - "pct_cuda_time": 0.618386867159623, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 514.602, - "cuda_time_us": 206.398, - "pct_cuda_time": 2.890323888856447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.326, - "cuda_time_us": 3.551, - "pct_cuda_time": 0.04972693596512197, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.551, - "pct_cuda_time": 0.04972693596512197, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 361.44, - "cuda_time_us": 67.744, - "pct_cuda_time": 0.9486627851369256, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.703, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.28813895646813564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.28813895646813564, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.348, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.050637172754120253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.050637172754120253, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.43, - "cuda_time_us": 25.855999999999998, - "pct_cuda_time": 0.3620781910206121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.808, - "pct_cuda_time": 0.2773841587150481, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018372779494857792, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.853, - "cuda_time_us": 17.695999999999998, - "pct_cuda_time": 0.2478084648940575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.21868088764611227, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.02912757724794528, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.487, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04570789045062182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04570789045062182, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 105.594, - "cuda_time_us": 131.839, - "pct_cuda_time": 1.8462262773037779, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.832, - "cuda_time_us": 79.168, - "pct_cuda_time": 1.108640401714102, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.168, - "pct_cuda_time": 1.108640401714102, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.756, - "cuda_time_us": 8.863, - "pct_cuda_time": 0.12411428709064376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.863, - "pct_cuda_time": 0.12411428709064376, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 37.564, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6134715884990322, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6134715884990322, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 486.246, - "cuda_time_us": 208.481, - "pct_cuda_time": 2.9194934770331153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.84, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04705224016975776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04705224016975776, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 344.358, - "cuda_time_us": 68.86500000000001, - "pct_cuda_time": 0.9643608688364194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.771, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.29217200562554346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.29217200562554346, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.703, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.052429639046301504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.052429639046301504, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 153.814, - "cuda_time_us": 26.337, - "pct_cuda_time": 0.3688139432591994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.031368160113171846, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 20.128, - "pct_cuda_time": 0.2818653244455013, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.02107548257603733, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.758, - "cuda_time_us": 17.92, - "pct_cuda_time": 0.25094528090537477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.808, - "pct_cuda_time": 0.22136958708438415, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.649, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04436354073148589, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.239, - "cuda_time_us": 133.088, - "pct_cuda_time": 1.8637168272954525, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.088, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1243244817706877, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1243244817706877, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.453, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1227839410144155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1227839410144155, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.609, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.6166084045103493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.6166084045103493, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 544.38, - "cuda_time_us": 207.70999999999998, - "pct_cuda_time": 2.908696668351305, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.636, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04705224016975776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04705224016975776, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 389.617, - "cuda_time_us": 68.704, - "pct_cuda_time": 0.9621062823282848, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.329, - "cuda_time_us": 20.96, - "pct_cuda_time": 0.2935163553446794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.96, - "pct_cuda_time": 0.2935163553446794, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.036, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05153340590021088, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 171.529, - "cuda_time_us": 26.112, - "pct_cuda_time": 0.36566312360497455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03181627668621715, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.872, - "pct_cuda_time": 0.2782803918611388, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021061478933129665, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 55.006, - "cuda_time_us": 17.951999999999998, - "pct_cuda_time": 0.25139339747842004, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.84, - "pct_cuda_time": 0.22181770365742945, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.607, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04570789045062182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.04570789045062182, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 106.951, - "cuda_time_us": 132.382, - "pct_cuda_time": 1.8538302554026405, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.391, - "cuda_time_us": 80.447, - "pct_cuda_time": 1.1265510609930067, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.447, - "pct_cuda_time": 1.1265510609930067, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 27.942, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12502452387964208, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12502452387964208, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.959, - "cuda_time_us": 43.007, - "pct_cuda_time": 0.6022546705299917, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.007, - "pct_cuda_time": 0.6022546705299917, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 501.007, - "cuda_time_us": 208.478, - "pct_cuda_time": 2.9194514661043924, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.756, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.04750035674280307, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 360.962, - "cuda_time_us": 69.086, - "pct_cuda_time": 0.9674556739190133, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.409, - "cuda_time_us": 21.76, - "pct_cuda_time": 0.3047192696708122, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.76, - "pct_cuda_time": 0.3047192696708122, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 107.771, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.050637172754120253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.050637172754120253, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.995, - "cuda_time_us": 25.982999999999997, - "pct_cuda_time": 0.3638566536698857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.305, - "pct_cuda_time": 0.03227839690217014, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.935, - "pct_cuda_time": 0.2791626213643217, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03405685955144371, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.311, - "pct_cuda_time": 0.018358775851950126, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.429, - "cuda_time_us": 17.727, - "pct_cuda_time": 0.2482425778241952, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.615, - "pct_cuda_time": 0.2186668840032046, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0295756938209906, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.576, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.043915424158440575, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.385, - "cuda_time_us": 132.864, - "pct_cuda_time": 1.8605800112841357, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.454, - "cuda_time_us": 81.504, - "pct_cuda_time": 1.1413529115464096, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.504, - "pct_cuda_time": 1.1413529115464096, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 19.678, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.12054335814918894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.12054335814918894, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.525, - "cuda_time_us": 42.752, - "pct_cuda_time": 0.5986837415885369, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.752, - "pct_cuda_time": 0.5986837415885369, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 496.01, - "cuda_time_us": 208.22199999999998, - "pct_cuda_time": 2.91586653352003, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.701, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04929282303498433, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04929282303498433, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 347.441, - "cuda_time_us": 68.733, - "pct_cuda_time": 0.9625123879726073, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.234, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.28724272332204503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.28724272332204503, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.13, - "cuda_time_us": 3.679, - "pct_cuda_time": 0.05151940225730322, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.679, - "pct_cuda_time": 0.05151940225730322, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.836, - "cuda_time_us": 26.08, - "pct_cuda_time": 0.3652150070319293, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03226439325926246, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 19.968, - "pct_cuda_time": 0.27962474158027467, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.018820896067903103, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 97], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.006, - "cuda_time_us": 18.462, - "pct_cuda_time": 0.2585352553613297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.351, - "pct_cuda_time": 0.22897356518324674, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.029561690178082932, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.879, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.045273777520484185, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.045273777520484185, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.742, - "cuda_time_us": 132.736, - "pct_cuda_time": 1.8587875449919542, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 39.225, - "cuda_time_us": 80.224, - "pct_cuda_time": 1.1234282486245972, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.224, - "pct_cuda_time": 1.1234282486245972, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.6, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1227839410144155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1227839410144155, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.053, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.6125753553529415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.6125753553529415, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.368, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0483965898888937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.0483965898888937, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 101.292, - "cuda_time_us": 349.279, - "pct_cuda_time": 4.891178391146673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.072, - "pct_cuda_time": 0.04301919101234995, - "trace": "index_select(bfloat16[2, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.010754797753087488, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 345.439, - "pct_cuda_time": 4.837404402381236, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 740.129, - "cuda_time_us": 123.746, - "pct_cuda_time": 1.7328947952520368, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03047192696708122, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03092004354012653, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.031368160113171846, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03092004354012653, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03092004354012653, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03092004354012653, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03047192696708122, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.16, - "pct_cuda_time": 0.05825515449589056, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.577, - "pct_cuda_time": 0.06409467358838727, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.271, - "pct_cuda_time": 0.4799188460886215, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.936, - "pct_cuda_time": 0.3912057682685574, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 1.825, - "pct_cuda_time": 0.025556648306490452, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.833, - "pct_cuda_time": 0.06767960617274978, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 28.256, - "pct_cuda_time": 0.39568693399901056, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03450497612448902, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - } -} \ No newline at end of file diff --git a/H100_llama8b_pp1_tp1/profiling_bs2_pl2048.json b/H100_llama8b_pp1_tp1/profiling_bs2_pl2048.json deleted file mode 100644 index bf82d077904a6c53258f1d37bb0fc3edffbb7e31..0000000000000000000000000000000000000000 --- a/H100_llama8b_pp1_tp1/profiling_bs2_pl2048.json +++ /dev/null @@ -1,18877 +0,0 @@ -{ - "context": { - "python_version": "3.12.9 | packaged by Anaconda, Inc. | (main, Feb 6 2025, 18:56:27) [GCC 11.2.0]", - "torch_version": "2.5.1+cu124", - "engine_args": { - "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "served_model_name": null, - "tokenizer": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "task": "auto", - "skip_tokenizer_init": false, - "tokenizer_mode": "auto", - "trust_remote_code": false, - "allowed_local_media_path": null, - "download_dir": null, - "load_format": "dummy", - "config_format": "auto", - "dtype": "auto", - "kv_cache_dtype": "auto", - "seed": 0, - "max_model_len": null, - "distributed_executor_backend": null, - "pipeline_parallel_size": 1, - "tensor_parallel_size": 1, - "max_parallel_loading_workers": null, - "block_size": null, - "enable_prefix_caching": false, - "disable_sliding_window": false, - "use_v2_block_manager": true, - "swap_space": 4, - "cpu_offload_gb": 0, - "gpu_memory_utilization": 0.9, - "max_num_batched_tokens": 8000, - "max_num_partial_prefills": 1, - "max_long_partial_prefills": 1, - "long_prefill_token_threshold": 0, - "max_num_seqs": 256, - "max_logprobs": 20, - "disable_log_stats": false, - "revision": null, - "code_revision": null, - "rope_scaling": null, - "rope_theta": null, - "hf_overrides": null, - "tokenizer_revision": null, - "quantization": null, - "enforce_eager": true, - "max_seq_len_to_capture": 8192, - "disable_custom_all_reduce": false, - "tokenizer_pool_size": 0, - "tokenizer_pool_type": "ray", - "tokenizer_pool_extra_config": null, - "limit_mm_per_prompt": null, - "mm_processor_kwargs": null, - "disable_mm_preprocessor_cache": false, - "enable_lora": false, - "enable_lora_bias": false, - "max_loras": 1, - "max_lora_rank": 16, - "enable_prompt_adapter": false, - "max_prompt_adapters": 1, - "max_prompt_adapter_token": 0, - "fully_sharded_loras": false, - "lora_extra_vocab_size": 256, - "long_lora_scaling_factors": null, - "lora_dtype": "auto", - "max_cpu_loras": null, - "device": "auto", - "num_scheduler_steps": 1, - "multi_step_stream_outputs": true, - "ray_workers_use_nsight": false, - "num_gpu_blocks_override": null, - "num_lookahead_slots": 0, - "model_loader_extra_config": null, - "ignore_patterns": [], - "preemption_mode": null, - "scheduler_delay_factor": 0.0, - "enable_chunked_prefill": null, - "guided_decoding_backend": "xgrammar", - "logits_processor_pattern": null, - "speculative_model": null, - "speculative_model_quantization": null, - "speculative_draft_tensor_parallel_size": null, - "num_speculative_tokens": null, - "speculative_disable_mqa_scorer": false, - "speculative_max_model_len": null, - "speculative_disable_by_batch_size": null, - "ngram_prompt_lookup_max": null, - "ngram_prompt_lookup_min": null, - "spec_decoding_acceptance_method": "rejection_sampler", - "typical_acceptance_sampler_posterior_threshold": null, - "typical_acceptance_sampler_posterior_alpha": null, - "qlora_adapter_name_or_path": null, - "disable_logprobs_during_spec_decoding": null, - "otlp_traces_endpoint": null, - "collect_detailed_traces": null, - "disable_async_output_proc": false, - "scheduling_policy": "fcfs", - "scheduler_cls": "vllm.core.scheduler.Scheduler", - "override_neuron_config": null, - "override_pooler_config": null, - "compilation_config": null, - "worker_cls": "auto", - "kv_transfer_config": null, - "generation_config": null, - "override_generation_config": null, - "enable_sleep_mode": false, - "model_impl": "auto", - "calculate_kv_scales": false, - "additional_config": null - }, - "prompt_len": 0, - "batch_size": 2, - "num_steps": 2, - "complete_num_requests_per_step": null, - "save_chrome_traces_folder": null - }, - "prefill": { - "metadata": { - "num_running_seqs": null - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 95224.18999999999, - "pct_cuda_time": 99.47766599017449, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 126.207, - "pct_cuda_time": 0.13184441675609895, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cuda_time_us": 126.207, - "pct_cuda_time": 0.13184441675609895, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 95052.159, - "pct_cuda_time": 99.2979507060859, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 2874.4590000000007, - "pct_cuda_time": 3.0028553910981133, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 70.56, - "pct_cuda_time": 0.07371177546657748, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 2803.899000000001, - "pct_cuda_time": 2.9291436156315362, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 23375.501, - "pct_cuda_time": 24.419638338020935, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 9201.707, - "pct_cuda_time": 9.612729029099125, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 32.00300000000001, - "pct_cuda_time": 0.033432510632892286, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 9169.704000000002, - "pct_cuda_time": 9.579296518466233, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 1805.597, - "pct_cuda_time": 1.886249442277861, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 1805.597, - "pct_cuda_time": 1.886249442277861, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 5877.809999999999, - "pct_cuda_time": 6.140360132585085, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 747.8709999999999, - "pct_cuda_time": 0.7812769165244439, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 5079.187999999997, - "pct_cuda_time": 5.30606526939533, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 50.75100000000001, - "pct_cuda_time": 0.05301794666531001, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 6490.387, - "pct_cuda_time": 6.780299734058863, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 31.615000000000013, - "pct_cuda_time": 0.033027179441267686, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 6458.771999999999, - "pct_cuda_time": 6.747272554617595, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 68802.199, - "pct_cuda_time": 71.87545697696684, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 43412.972, - "pct_cuda_time": 45.35214348640611, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 31.774000000000008, - "pct_cuda_time": 0.033193281656392196, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 43381.198, - "pct_cuda_time": 45.31895020474971, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 5750.642000000001, - "pct_cuda_time": 6.007511789862105, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 5750.642000000001, - "pct_cuda_time": 6.007511789862105, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 19638.585000000003, - "pct_cuda_time": 20.51580170069865, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 30.75300000000001, - "pct_cuda_time": 0.032126675608328485, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 19607.832000000002, - "pct_cuda_time": 20.48367502509032, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 45.824, - "pct_cuda_time": 0.04787086733248931, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 45.824, - "pct_cuda_time": 0.04787086733248931, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 360.288, - "pct_cuda_time": 0.3763813514640344, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.872, - "pct_cuda_time": 0.004044954572088832, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 355.36, - "pct_cuda_time": 0.37123322746319404, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 139.71200000000002, - "pct_cuda_time": 0.1459526583614863, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 16.192, - "pct_cuda_time": 0.01691526457418966, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.448, - "pct_cuda_time": 0.004646683351407832, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 5.248, - "pct_cuda_time": 0.005482417767128665, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 39.52, - "pct_cuda_time": 0.041285280136609155, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 31.744, - "pct_cuda_time": 0.03316194161580266, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 2.176, - "pct_cuda_time": 0.0022731976107606662, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 5.312, - "pct_cuda_time": 0.005549276520386332, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 31.904, - "pct_cuda_time": 0.03332908849894682, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 3.168, - "pct_cuda_time": 0.0033095082862544993, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 17945.741, - "cuda_time_us": 95224.18999999999, - "pct_cuda_time": 99.47766599017449, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 63.818, - "cuda_time_us": 126.207, - "pct_cuda_time": 0.13184441675609895, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 126.207, - "pct_cuda_time": 0.13184441675609895, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[4096]) <- embedding(bfloat16[128256, 4096], int64[4096], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 853.481, - "cuda_time_us": 3007.321, - "pct_cuda_time": 3.1416520735249893, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 48.764, - "cuda_time_us": 70.56, - "pct_cuda_time": 0.07371177546657748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 70.56, - "pct_cuda_time": 0.07371177546657748, - "trace": "_C::rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 626.016, - "cuda_time_us": 735.742, - "pct_cuda_time": 0.7686061381140964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 85.936, - "cuda_time_us": 290.07899999999995, - "pct_cuda_time": 0.3030362544723544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 289.311, - "pct_cuda_time": 0.3022339494332624, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 194.494, - "cuda_time_us": 56.64, - "pct_cuda_time": 0.05916999663303498, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.64, - "pct_cuda_time": 0.05916999663303498, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 222.809, - "cuda_time_us": 186.33599999999998, - "pct_cuda_time": 0.19465926010969642, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.392, - "pct_cuda_time": 0.02443687431567716, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 161.536, - "pct_cuda_time": 0.1687514932223506, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.001470892571668666, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.574, - "cuda_time_us": 202.687, - "pct_cuda_time": 0.21174062689901063, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 201.919, - "pct_cuda_time": 0.21093832185991862, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.044, - "cuda_time_us": 44.384, - "pct_cuda_time": 0.04636654538419182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.384, - "pct_cuda_time": 0.04636654538419182, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 126.114, - "cuda_time_us": 2156.635, - "pct_cuda_time": 2.2529676145601236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 43.495, - "cuda_time_us": 1364.54, - "pct_cuda_time": 1.4254912995346318, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007678309944435153, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1363.805, - "pct_cuda_time": 1.4247234685401886, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 29.153, - "cuda_time_us": 180.064, - "pct_cuda_time": 0.18810710229044508, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.064, - "pct_cuda_time": 0.18810710229044508, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 37.515, - "cuda_time_us": 612.031, - "pct_cuda_time": 0.6393692127350464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 611.295, - "pct_cuda_time": 0.6386003370725832, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 559.07, - "cuda_time_us": 2975.417, - "pct_cuda_time": 3.1083229850260423, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.829, - "cuda_time_us": 45.184, - "pct_cuda_time": 0.04720227979991265, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.184, - "pct_cuda_time": 0.04720227979991265, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 396.058, - "cuda_time_us": 732.03, - "pct_cuda_time": 0.7647283304251518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.028, - "cuda_time_us": 287.519, - "pct_cuda_time": 0.30036190434204774, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011365988053803331, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 286.431, - "pct_cuda_time": 0.29922530553666743, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 115.844, - "cuda_time_us": 56.608, - "pct_cuda_time": 0.05913656725640615, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.608, - "pct_cuda_time": 0.05913656725640615, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 167.662, - "cuda_time_us": 184.54299999999998, - "pct_cuda_time": 0.1927861703504621, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 24.064, - "pct_cuda_time": 0.025138891224882658, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 158.879, - "pct_cuda_time": 0.16597581029413777, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.0016714688314416663, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.918, - "cuda_time_us": 203.35999999999999, - "pct_cuda_time": 0.21244368847623574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.001470892571668666, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 201.952, - "pct_cuda_time": 0.2109727959045671, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.178, - "cuda_time_us": 44.576, - "pct_cuda_time": 0.046567121643964815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.576, - "pct_cuda_time": 0.046567121643964815, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 111.873, - "cuda_time_us": 2153.627, - "pct_cuda_time": 2.249825253157013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 45.125, - "cuda_time_us": 1360.765, - "pct_cuda_time": 1.4215476777604494, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1359.997, - "pct_cuda_time": 1.4207453727213573, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.623, - "cuda_time_us": 180.319, - "pct_cuda_time": 0.1883734926354561, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.319, - "pct_cuda_time": 0.1883734926354561, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.869, - "cuda_time_us": 612.543, - "pct_cuda_time": 0.6399040827611078, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015043219482974994, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 611.103, - "pct_cuda_time": 0.6383997608128102, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 515.884, - "cuda_time_us": 2977.404, - "pct_cuda_time": 3.110398740381089, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.949, - "cuda_time_us": 44.96, - "pct_cuda_time": 0.04696827416351081, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.96, - "pct_cuda_time": 0.04696827416351081, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 367.316, - "cuda_time_us": 732.638, - "pct_cuda_time": 0.7653634885810996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.136, - "cuda_time_us": 286.783, - "pct_cuda_time": 0.2995930286795846, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.184, - "pct_cuda_time": 0.001236886935266833, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 285.599, - "pct_cuda_time": 0.29835614174431774, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.934, - "cuda_time_us": 56.672, - "pct_cuda_time": 0.05920342600966381, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.672, - "pct_cuda_time": 0.05920342600966381, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.218, - "cuda_time_us": 185.47199999999998, - "pct_cuda_time": 0.19375666694071791, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.296, - "pct_cuda_time": 0.024336586185790657, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 160.128, - "pct_cuda_time": 0.16728060065068193, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 2.048, - "pct_cuda_time": 0.0021394801042453324, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.791, - "cuda_time_us": 203.71099999999998, - "pct_cuda_time": 0.21281036695113328, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 202.975, - "pct_cuda_time": 0.2120414912886701, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.64, - "cuda_time_us": 45.025, - "pct_cuda_time": 0.04703617758478813, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.025, - "pct_cuda_time": 0.04703617758478813, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.559, - "cuda_time_us": 2154.781, - "pct_cuda_time": 2.2510308000516908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.485, - "cuda_time_us": 1361.183, - "pct_cuda_time": 1.4219843489926633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.089, - "pct_cuda_time": 0.001137643473399984, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1360.094, - "pct_cuda_time": 1.4208467055192635, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.615, - "cuda_time_us": 180.799, - "pct_cuda_time": 0.1888749332848886, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.799, - "pct_cuda_time": 0.1888749332848886, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.357, - "cuda_time_us": 612.799, - "pct_cuda_time": 0.6401715177741385, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011700281820091665, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 611.679, - "pct_cuda_time": 0.6390014895921293, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 556.261, - "cuda_time_us": 2911.0989999999997, - "pct_cuda_time": 3.0411320273381266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.264, - "cuda_time_us": 45.632, - "pct_cuda_time": 0.047670291072716316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.632, - "pct_cuda_time": 0.047670291072716316, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 394.158, - "cuda_time_us": 734.366, - "pct_cuda_time": 0.7671686749190566, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.115, - "cuda_time_us": 287.48699999999997, - "pct_cuda_time": 0.30032847496541887, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 286.719, - "pct_cuda_time": 0.2995261699263269, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.085, - "cuda_time_us": 56.543, - "pct_cuda_time": 0.05906866383512883, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.543, - "pct_cuda_time": 0.05906866383512883, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.125, - "cuda_time_us": 186.208, - "pct_cuda_time": 0.1945255426031811, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.84, - "pct_cuda_time": 0.024904885588480825, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 160.832, - "pct_cuda_time": 0.16801604693651626, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0016046100781839994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.683, - "cuda_time_us": 204.128, - "pct_cuda_time": 0.21324599351532775, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.696, - "pct_cuda_time": 0.001771756961328166, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 202.432, - "pct_cuda_time": 0.2114742365539996, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.29, - "cuda_time_us": 44.288, - "pct_cuda_time": 0.04626625725430531, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.288, - "pct_cuda_time": 0.04626625725430531, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 112.142, - "cuda_time_us": 2086.8129999999996, - "pct_cuda_time": 2.180026804092048, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 39.271, - "cuda_time_us": 1326.3339999999998, - "pct_cuda_time": 1.3855787131758441, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.696, - "pct_cuda_time": 0.001771756961328166, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1324.638, - "pct_cuda_time": 1.383806956214516, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 24.943, - "cuda_time_us": 176.511, - "pct_cuda_time": 0.18439539681662495, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 176.511, - "pct_cuda_time": 0.18439539681662495, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 35.349, - "cuda_time_us": 583.968, - "pct_cuda_time": 0.6100526940995793, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0007699203304828174, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 583.231, - "pct_cuda_time": 0.6092827737690965, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 539.286, - "cuda_time_us": 2807.77, - "pct_cuda_time": 2.933187525535604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.134, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.04586510473475932, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.904, - "pct_cuda_time": 0.04586510473475932, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 385.269, - "cuda_time_us": 692.99, - "pct_cuda_time": 0.7239444909379752, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 35.201, - "cuda_time_us": 275.135, - "pct_cuda_time": 0.28742473558668924, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015043219482974994, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 273.695, - "pct_cuda_time": 0.2859204136383917, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 107.799, - "cuda_time_us": 54.752, - "pct_cuda_time": 0.057197663411933816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.752, - "pct_cuda_time": 0.057197663411933816, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.933, - "cuda_time_us": 171.904, - "pct_cuda_time": 0.17958261125009262, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.944, - "pct_cuda_time": 0.02396886304287349, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 147.584, - "pct_cuda_time": 0.15417628501217928, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.0014374631950398327, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.586, - "cuda_time_us": 191.199, - "pct_cuda_time": 0.19973948068925948, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 190.431, - "pct_cuda_time": 0.19893717565016744, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.929, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.04599882224127465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.04599882224127465, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 103.016, - "cuda_time_us": 2026.844, - "pct_cuda_time": 2.1173791076215953, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.622, - "cuda_time_us": 1265.597, - "pct_cuda_time": 1.322128711666299, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1264.541, - "pct_cuda_time": 1.3210255422375474, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.737, - "cuda_time_us": 176.096, - "pct_cuda_time": 0.18396185958846978, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 176.096, - "pct_cuda_time": 0.18396185958846978, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.47, - "cuda_time_us": 585.151, - "pct_cuda_time": 0.6112885363668265, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 584.415, - "pct_cuda_time": 0.6105196607043633, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 589.938, - "cuda_time_us": 2806.588, - "pct_cuda_time": 2.931952727936377, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.784, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.04599882224127465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.04599882224127465, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 423.324, - "cuda_time_us": 691.168, - "pct_cuda_time": 0.722041105806171, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.644, - "cuda_time_us": 272.896, - "pct_cuda_time": 0.2850857238906906, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 272.16, - "pct_cuda_time": 0.2843168482282274, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 140.963, - "cuda_time_us": 54.592, - "pct_cuda_time": 0.05703051652878965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.592, - "pct_cuda_time": 0.05703051652878965, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.229, - "cuda_time_us": 172.38299999999998, - "pct_cuda_time": 0.18008300723150544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.592, - "pct_cuda_time": 0.023601139899956325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 148.255, - "pct_cuda_time": 0.15487725725336512, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0016046100781839994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 53.197, - "cuda_time_us": 191.297, - "pct_cuda_time": 0.19984185815518524, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.377, - "pct_cuda_time": 0.0014385078630594838, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 189.92, - "pct_cuda_time": 0.19840335029212577, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.495, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.04599882224127465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.04599882224127465, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 106.39, - "cuda_time_us": 2027.3560000000002, - "pct_cuda_time": 2.1179139776476568, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 40.331, - "cuda_time_us": 1265.117, - "pct_cuda_time": 1.3216272710168664, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1264.349, - "pct_cuda_time": 1.3208249659777742, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.812, - "cuda_time_us": 176.256, - "pct_cuda_time": 0.18412900647161393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 176.256, - "pct_cuda_time": 0.18412900647161393, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.785, - "cuda_time_us": 585.9830000000001, - "pct_cuda_time": 0.6121577001591761, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.0014040338184109996, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 584.639, - "pct_cuda_time": 0.6107536663407652, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 512.513, - "cuda_time_us": 2808.921, - "pct_cuda_time": 2.9343899384262224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.657, - "cuda_time_us": 44.256, - "pct_cuda_time": 0.046232827877676484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.256, - "pct_cuda_time": 0.046232827877676484, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 364.027, - "cuda_time_us": 692.638, - "pct_cuda_time": 0.723576767795058, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.521, - "cuda_time_us": 274.207, - "pct_cuda_time": 0.28645528366445305, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.248, - "pct_cuda_time": 0.0013037456885244996, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 272.959, - "pct_cuda_time": 0.28515153797592857, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.096, - "cuda_time_us": 54.752, - "pct_cuda_time": 0.057197663411933816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.752, - "pct_cuda_time": 0.057197663411933816, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.469, - "cuda_time_us": 172.096, - "pct_cuda_time": 0.1797831875098656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.816, - "pct_cuda_time": 0.023835145536358158, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 147.744, - "pct_cuda_time": 0.15434343189532346, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0016046100781839994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 45.155, - "cuda_time_us": 191.583, - "pct_cuda_time": 0.20014063320880543, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 190.815, - "pct_cuda_time": 0.19933832816971345, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.096, - "cuda_time_us": 43.807, - "pct_cuda_time": 0.04576377193685317, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.807, - "pct_cuda_time": 0.04576377193685317, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.243, - "cuda_time_us": 2028.2199999999998, - "pct_cuda_time": 2.1188165708166347, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.635, - "cuda_time_us": 1267.07, - "pct_cuda_time": 1.323667507659245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.248, - "pct_cuda_time": 0.0013037456885244996, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1265.822, - "pct_cuda_time": 1.3223637619707205, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.642, - "cuda_time_us": 176.223, - "pct_cuda_time": 0.18409453242696547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 176.223, - "pct_cuda_time": 0.18409453242696547, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.092, - "cuda_time_us": 584.927, - "pct_cuda_time": 0.6110545307304247, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0007699203304828174, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 584.19, - "pct_cuda_time": 0.6102846103999419, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 531.733, - "cuda_time_us": 2806.938, - "pct_cuda_time": 2.932318361743255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.053, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.04593196348801699, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.968, - "pct_cuda_time": 0.04593196348801699, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 386.22, - "cuda_time_us": 691.5819999999999, - "pct_cuda_time": 0.7224735983663064, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.011, - "cuda_time_us": 273.343, - "pct_cuda_time": 0.2855526904954746, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 272.319, - "pct_cuda_time": 0.2844829504433519, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 116.105, - "cuda_time_us": 54.816, - "pct_cuda_time": 0.057264522165191484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.816, - "pct_cuda_time": 0.057264522165191484, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 162.442, - "cuda_time_us": 171.58399999999997, - "pct_cuda_time": 0.17924831748380426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.04, - "pct_cuda_time": 0.02406915117275999, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 147.2, - "pct_cuda_time": 0.15377513249263328, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.0014040338184109996, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.606, - "cuda_time_us": 191.839, - "pct_cuda_time": 0.20040806822183613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 190.783, - "pct_cuda_time": 0.1993048987930846, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.265, - "cuda_time_us": 43.264, - "pct_cuda_time": 0.04519651720218266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.264, - "pct_cuda_time": 0.04519651720218266, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.48, - "cuda_time_us": 2028.124, - "pct_cuda_time": 2.1187162826867487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.978, - "cuda_time_us": 1266.717, - "pct_cuda_time": 1.3232987398483083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.0014029891503913486, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1265.374, - "pct_cuda_time": 1.3218957506979168, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.288, - "cuda_time_us": 175.904, - "pct_cuda_time": 0.18376128332869676, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 175.904, - "pct_cuda_time": 0.18376128332869676, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.188, - "cuda_time_us": 585.5029999999999, - "pct_cuda_time": 0.6116562595097436, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.025, - "pct_cuda_time": 0.0010707847201423173, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 584.478, - "pct_cuda_time": 0.6105854747896012, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 529.066, - "cuda_time_us": 2809.369, - "pct_cuda_time": 2.9348579496990266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.19, - "cuda_time_us": 44.256, - "pct_cuda_time": 0.046232827877676484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.256, - "pct_cuda_time": 0.046232827877676484, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 359.743, - "cuda_time_us": 691.069, - "pct_cuda_time": 0.7219376836722254, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 36.586, - "cuda_time_us": 273.311, - "pct_cuda_time": 0.28551926111884574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 272.287, - "pct_cuda_time": 0.28444952106672305, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.791, - "cuda_time_us": 54.751, - "pct_cuda_time": 0.05719661874391416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.751, - "pct_cuda_time": 0.05719661874391416, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.744, - "cuda_time_us": 171.327, - "pct_cuda_time": 0.17897983780275395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.495, - "pct_cuda_time": 0.023499807102050175, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 147.296, - "pct_cuda_time": 0.15387542062251977, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0016046100781839994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.494, - "cuda_time_us": 191.67999999999998, - "pct_cuda_time": 0.20024196600671157, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 190.944, - "pct_cuda_time": 0.19947309034424843, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.269, - "cuda_time_us": 43.936, - "pct_cuda_time": 0.04589853411138815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.936, - "pct_cuda_time": 0.04589853411138815, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 123.785, - "cuda_time_us": 2030.1080000000002, - "pct_cuda_time": 2.1207889040377363, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.941, - "cuda_time_us": 1270.077, - "pct_cuda_time": 1.3268088243943357, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011700281820091665, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1268.957, - "pct_cuda_time": 1.3256387962123266, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.395, - "cuda_time_us": 175.872, - "pct_cuda_time": 0.18372785395206795, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 175.872, - "pct_cuda_time": 0.18372785395206795, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 58.629, - "cuda_time_us": 584.159, - "pct_cuda_time": 0.6102522256913326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 583.391, - "pct_cuda_time": 0.6094499206522406, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 519.736, - "cuda_time_us": 2809.8830000000003, - "pct_cuda_time": 2.9353949090611273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.626, - "cuda_time_us": 45.152, - "pct_cuda_time": 0.04716885042328382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.152, - "pct_cuda_time": 0.04716885042328382, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 372.73, - "cuda_time_us": 692.223, - "pct_cuda_time": 0.7231432305669027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 37.99, - "cuda_time_us": 273.98299999999995, - "pct_cuda_time": 0.2862212780280512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 273.215, - "pct_cuda_time": 0.28541897298895924, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.319, - "cuda_time_us": 54.784, - "pct_cuda_time": 0.05723109278856264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.784, - "pct_cuda_time": 0.05723109278856264, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.237, - "cuda_time_us": 171.328, - "pct_cuda_time": 0.1789808824707736, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.784, - "pct_cuda_time": 0.023801716159729324, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 146.848, - "pct_cuda_time": 0.15340740934971614, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.696, - "pct_cuda_time": 0.001771756961328166, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.338, - "cuda_time_us": 192.128, - "pct_cuda_time": 0.20070997727951526, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011365988053803331, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 191.04, - "pct_cuda_time": 0.19957337847413492, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.396, - "cuda_time_us": 42.688, - "pct_cuda_time": 0.04459478842286365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 42.688, - "pct_cuda_time": 0.04459478842286365, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.321, - "cuda_time_us": 2029.8200000000002, - "pct_cuda_time": 2.1204880396480768, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.185, - "cuda_time_us": 1267.9660000000001, - "pct_cuda_time": 1.3246035302048522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1267.198, - "pct_cuda_time": 1.3238012251657603, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.734, - "cuda_time_us": 176.064, - "pct_cuda_time": 0.18392843021184094, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 176.064, - "pct_cuda_time": 0.18392843021184094, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.636, - "cuda_time_us": 585.7900000000001, - "pct_cuda_time": 0.6119560792313836, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.248, - "pct_cuda_time": 0.0013037456885244996, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 584.542, - "pct_cuda_time": 0.6106523335428591, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 510.754, - "cuda_time_us": 2812.152, - "pct_cuda_time": 2.937765260797715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.1, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.046466833514078314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.046466833514078314, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 368.857, - "cuda_time_us": 691.422, - "pct_cuda_time": 0.7223064514831623, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.043, - "cuda_time_us": 274.495, - "pct_cuda_time": 0.2867561480541126, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.001036310675493833, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 273.503, - "pct_cuda_time": 0.2857198373786187, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.863, - "cuda_time_us": 54.336, - "pct_cuda_time": 0.056763081515758984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 54.336, - "pct_cuda_time": 0.056763081515758984, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.911, - "cuda_time_us": 171.2, - "pct_cuda_time": 0.17884716496425826, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.02363456927658516, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 147.04, - "pct_cuda_time": 0.1536079856094891, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0016046100781839994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.245, - "cuda_time_us": 191.391, - "pct_cuda_time": 0.19994005694903244, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 190.655, - "pct_cuda_time": 0.1991711812865693, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.522, - "cuda_time_us": 43.839, - "pct_cuda_time": 0.045797201313482, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.839, - "pct_cuda_time": 0.045797201313482, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.311, - "cuda_time_us": 2032.411, - "pct_cuda_time": 2.1231947744869926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.272, - "cuda_time_us": 1267.966, - "pct_cuda_time": 1.3246035302048522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1266.942, - "pct_cuda_time": 1.3235337901527295, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.262, - "cuda_time_us": 176.063, - "pct_cuda_time": 0.18392738554382126, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 176.063, - "pct_cuda_time": 0.18392738554382126, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.729, - "cuda_time_us": 588.3820000000001, - "pct_cuda_time": 0.6146638587383191, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 587.614, - "pct_cuda_time": 0.613861553699227, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 559.339, - "cuda_time_us": 2926.3280000000004, - "pct_cuda_time": 3.057041276609393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.802, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.046466833514078314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.046466833514078314, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 401.058, - "cuda_time_us": 709.885, - "pct_cuda_time": 0.7415941571299794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.281, - "cuda_time_us": 278.84700000000004, - "pct_cuda_time": 0.29130254327563393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.992, - "pct_cuda_time": 0.001036310675493833, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 277.855, - "pct_cuda_time": 0.2902662326001401, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.835, - "cuda_time_us": 55.551, - "pct_cuda_time": 0.058032353159634995, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 55.551, - "pct_cuda_time": 0.058032353159634995, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 162.446, - "cuda_time_us": 178.271, - "pct_cuda_time": 0.18623401253121077, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.023902004289615825, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 153.855, - "pct_cuda_time": 0.16072739816341094, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0016046100781839994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.64, - "cuda_time_us": 197.216, - "pct_cuda_time": 0.20602524816349976, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 196.448, - "pct_cuda_time": 0.20522294312440778, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.593, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.045664528474986324, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.045664528474986324, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 106.553, - "cuda_time_us": 2128.251, - "pct_cuda_time": 2.2233157574903486, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.586, - "cuda_time_us": 1333.436, - "pct_cuda_time": 1.3929979454514059, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007678309944435153, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1332.701, - "pct_cuda_time": 1.3922301144569624, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 23.792, - "cuda_time_us": 180.352, - "pct_cuda_time": 0.18840796668010462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.352, - "pct_cuda_time": 0.18840796668010462, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.537, - "cuda_time_us": 614.4630000000001, - "pct_cuda_time": 0.6419098453588379, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 613.407, - "pct_cuda_time": 0.6408066759300863, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 605.641, - "cuda_time_us": 2995.962, - "pct_cuda_time": 3.1297856894897733, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.465, - "cuda_time_us": 45.663, - "pct_cuda_time": 0.0477026757813255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.663, - "pct_cuda_time": 0.0477026757813255, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 453.728, - "cuda_time_us": 739.776, - "pct_cuda_time": 0.7728203289053687, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.058, - "cuda_time_us": 291.169, - "pct_cuda_time": 0.30417494261377404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.313, - "pct_cuda_time": 0.0013716491098018171, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 289.856, - "pct_cuda_time": 0.3028032935039722, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 114.226, - "cuda_time_us": 56.96, - "pct_cuda_time": 0.05950429039932332, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.96, - "pct_cuda_time": 0.05950429039932332, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 218.878, - "cuda_time_us": 186.719, - "pct_cuda_time": 0.19505936796122278, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.136, - "pct_cuda_time": 0.024169439302646492, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 162.207, - "pct_cuda_time": 0.16945246546353646, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.0014374631950398327, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.602, - "cuda_time_us": 204.928, - "pct_cuda_time": 0.2140817279310486, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 204.16, - "pct_cuda_time": 0.2132794228919566, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.546, - "cuda_time_us": 44.8, - "pct_cuda_time": 0.04680112728036665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.8, - "pct_cuda_time": 0.04680112728036665, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.344, - "cuda_time_us": 2165.723, - "pct_cuda_time": 2.262461557522712, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.261, - "cuda_time_us": 1364.733, - "pct_cuda_time": 1.4256929204624245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011365988053803331, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1363.645, - "pct_cuda_time": 1.424556321657044, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.024, - "cuda_time_us": 180.192, - "pct_cuda_time": 0.18824081979696045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.192, - "pct_cuda_time": 0.18824081979696045, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.757, - "cuda_time_us": 620.798, - "pct_cuda_time": 0.6485278172633271, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 620.062, - "pct_cuda_time": 0.647758941600864, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 512.396, - "cuda_time_us": 2991.548, - "pct_cuda_time": 3.1251745248510328, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.308, - "cuda_time_us": 45.152, - "pct_cuda_time": 0.04716885042328382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.152, - "pct_cuda_time": 0.04716885042328382, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 366.454, - "cuda_time_us": 734.751, - "pct_cuda_time": 0.7675708721066222, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.779, - "cuda_time_us": 289.08700000000005, - "pct_cuda_time": 0.3019999437968606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011365988053803331, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 287.999, - "pct_cuda_time": 0.30086334499148026, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.825, - "cuda_time_us": 56.128, - "pct_cuda_time": 0.05863512660697365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.128, - "pct_cuda_time": 0.05863512660697365, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 153.043, - "cuda_time_us": 184.64, - "pct_cuda_time": 0.19288750314836825, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.808, - "pct_cuda_time": 0.02487145621185199, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 159.232, - "pct_cuda_time": 0.1663445781050746, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.0016714688314416663, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.94, - "cuda_time_us": 204.896, - "pct_cuda_time": 0.21404829855441976, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.0014040338184109996, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 203.552, - "pct_cuda_time": 0.21264426473600875, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.637, - "cuda_time_us": 44.16, - "pct_cuda_time": 0.04613253974778998, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.16, - "pct_cuda_time": 0.04613253974778998, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.546, - "cuda_time_us": 2167.4849999999997, - "pct_cuda_time": 2.264302262573337, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.124, - "cuda_time_us": 1373.405, - "pct_cuda_time": 1.4347522815288383, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.767, - "pct_cuda_time": 0.0008012603710723488, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1372.638, - "pct_cuda_time": 1.433951021157766, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.221, - "cuda_time_us": 180.063, - "pct_cuda_time": 0.18810605762242544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.063, - "pct_cuda_time": 0.18810605762242544, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.424, - "cuda_time_us": 614.017, - "pct_cuda_time": 0.6414439234220735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.089, - "pct_cuda_time": 0.001137643473399984, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 612.928, - "pct_cuda_time": 0.6403062799486735, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 529.99, - "cuda_time_us": 2997.5009999999997, - "pct_cuda_time": 3.1313934335720153, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.89, - "cuda_time_us": 44.416, - "pct_cuda_time": 0.04639997476082065, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.416, - "pct_cuda_time": 0.04639997476082065, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 380.19, - "cuda_time_us": 735.617, - "pct_cuda_time": 0.76847555461164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.844, - "cuda_time_us": 288.864, - "pct_cuda_time": 0.30176698282847836, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0010028812988649995, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 287.904, - "pct_cuda_time": 0.30076410152961336, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.659, - "cuda_time_us": 56.577, - "pct_cuda_time": 0.059104182547796966, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.577, - "pct_cuda_time": 0.059104182547796966, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 168.483, - "cuda_time_us": 185.56799999999998, - "pct_cuda_time": 0.1938569550706044, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.616, - "pct_cuda_time": 0.024670879952078992, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 160.192, - "pct_cuda_time": 0.1673474594039396, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.76, - "pct_cuda_time": 0.0018386157145858328, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 49.038, - "cuda_time_us": 204.608, - "pct_cuda_time": 0.21374743416476025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011700281820091665, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 203.488, - "pct_cuda_time": 0.2125774059827511, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.676, - "cuda_time_us": 43.776, - "pct_cuda_time": 0.04573138722824399, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.776, - "pct_cuda_time": 0.04573138722824399, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.084, - "cuda_time_us": 2173.692, - "pct_cuda_time": 2.2707865169713113, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.622, - "cuda_time_us": 1373.8210000000001, - "pct_cuda_time": 1.4351868634250133, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1373.085, - "pct_cuda_time": 1.43441798776255, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.02, - "cuda_time_us": 179.328, - "pct_cuda_time": 0.18733822662798194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 179.328, - "pct_cuda_time": 0.18733822662798194, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.571, - "cuda_time_us": 620.543, - "pct_cuda_time": 0.6482614269183161, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 619.775, - "pct_cuda_time": 0.6474591218792242, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 519.103, - "cuda_time_us": 2985.401, - "pct_cuda_time": 3.118752950534238, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.46, - "cuda_time_us": 45.216, - "pct_cuda_time": 0.047235709176541484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.216, - "pct_cuda_time": 0.047235709176541484, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 368.748, - "cuda_time_us": 735.934, - "pct_cuda_time": 0.7688067143738695, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.352, - "cuda_time_us": 289.69599999999997, - "pct_cuda_time": 0.30263614662082805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0010028812988649995, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 288.736, - "pct_cuda_time": 0.30163326532196305, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.379, - "cuda_time_us": 56.736, - "pct_cuda_time": 0.05927028476292147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.736, - "pct_cuda_time": 0.05927028476292147, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.508, - "cuda_time_us": 184.894, - "pct_cuda_time": 0.19315284882535966, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.328, - "pct_cuda_time": 0.02437001556241949, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 160.159, - "pct_cuda_time": 0.16731298535929112, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.407, - "pct_cuda_time": 0.0014698479036490152, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.07, - "cuda_time_us": 204.608, - "pct_cuda_time": 0.21374743416476025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 203.872, - "pct_cuda_time": 0.2129785585022971, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.954, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.04576481660487282, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.04576481660487282, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.909, - "cuda_time_us": 2160.4429999999998, - "pct_cuda_time": 2.2569457103789543, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.929, - "cuda_time_us": 1363.293, - "pct_cuda_time": 1.424188598514127, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.001470892571668666, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1361.885, - "pct_cuda_time": 1.4227177059424583, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.77, - "cuda_time_us": 180.639, - "pct_cuda_time": 0.18870778640174446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.639, - "pct_cuda_time": 0.18870778640174446, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.123, - "cuda_time_us": 616.5110000000001, - "pct_cuda_time": 0.6440493254630832, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 615.455, - "pct_cuda_time": 0.6429461560343317, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 533.997, - "cuda_time_us": 3001.37, - "pct_cuda_time": 3.135435254140045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.811, - "cuda_time_us": 44.576, - "pct_cuda_time": 0.046567121643964815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.576, - "pct_cuda_time": 0.046567121643964815, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 383.272, - "cuda_time_us": 736.0629999999999, - "pct_cuda_time": 0.7689414765484043, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.944, - "cuda_time_us": 289.727, - "pct_cuda_time": 0.3026685313294372, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 288.703, - "pct_cuda_time": 0.30159879127731454, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.274, - "cuda_time_us": 56.832, - "pct_cuda_time": 0.059370572892807984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.832, - "pct_cuda_time": 0.059370572892807984, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 169.903, - "cuda_time_us": 184.70399999999998, - "pct_cuda_time": 0.19295436190162593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.296, - "pct_cuda_time": 0.024336586185790657, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 159.968, - "pct_cuda_time": 0.16711345376753778, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015043219482974994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.337, - "cuda_time_us": 204.8, - "pct_cuda_time": 0.2139480104245333, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 203.776, - "pct_cuda_time": 0.21287827037241058, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.004, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.04553081096847099, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.584, - "pct_cuda_time": 0.04553081096847099, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.081, - "cuda_time_us": 2177.147, - "pct_cuda_time": 2.2743958449792054, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.388, - "cuda_time_us": 1380.094, - "pct_cuda_time": 1.4417400659122843, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1379.358, - "pct_cuda_time": 1.440971190249821, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.079, - "cuda_time_us": 180.255, - "pct_cuda_time": 0.18830663388219845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.255, - "pct_cuda_time": 0.18830663388219845, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.747, - "cuda_time_us": 616.798, - "pct_cuda_time": 0.644349145184723, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007678309944435153, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 616.063, - "pct_cuda_time": 0.6435813141902794, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 505.746, - "cuda_time_us": 3013.34, - "pct_cuda_time": 3.147939930335269, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.891, - "cuda_time_us": 44.544, - "pct_cuda_time": 0.04653369226733598, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.544, - "pct_cuda_time": 0.04653369226733598, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 353.281, - "cuda_time_us": 739.232, - "pct_cuda_time": 0.7722520295026786, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.312, - "cuda_time_us": 290.71999999999997, - "pct_cuda_time": 0.30370588667295073, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0010028812988649995, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 289.76, - "pct_cuda_time": 0.30270300537408573, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.918, - "cuda_time_us": 56.672, - "pct_cuda_time": 0.05920342600966381, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.672, - "pct_cuda_time": 0.05920342600966381, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 151.979, - "cuda_time_us": 187.104, - "pct_cuda_time": 0.19546156514878843, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.488, - "pct_cuda_time": 0.02453716244556366, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 161.888, - "pct_cuda_time": 0.16911921636526778, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.728, - "pct_cuda_time": 0.0018051863379569993, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 39.878, - "cuda_time_us": 204.736, - "pct_cuda_time": 0.21388115167127558, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 204.0, - "pct_cuda_time": 0.21311227600881244, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.343, - "cuda_time_us": 43.457, - "pct_cuda_time": 0.045398138129975306, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.457, - "pct_cuda_time": 0.045398138129975306, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.06, - "cuda_time_us": 2186.107, - "pct_cuda_time": 2.2837560704352793, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.186, - "cuda_time_us": 1385.119, - "pct_cuda_time": 1.4469895227110305, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0011042140967711506, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1384.062, - "pct_cuda_time": 1.4458853086142593, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 23.523, - "cuda_time_us": 180.927, - "pct_cuda_time": 0.18900865079140394, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.927, - "pct_cuda_time": 0.18900865079140394, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.307, - "cuda_time_us": 620.0609999999999, - "pct_cuda_time": 0.6477578969328442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.0014029891503913486, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 618.718, - "pct_cuda_time": 0.646354907782453, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 518.439, - "cuda_time_us": 3014.745, - "pct_cuda_time": 3.1494076889028784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.163, - "cuda_time_us": 45.664, - "pct_cuda_time": 0.04770372044934515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.664, - "pct_cuda_time": 0.04770372044934515, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 356.144, - "cuda_time_us": 743.549, - "pct_cuda_time": 0.7767618613435121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.27, - "cuda_time_us": 292.191, - "pct_cuda_time": 0.30524259332985737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 291.167, - "pct_cuda_time": 0.3041728532777347, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.466, - "cuda_time_us": 56.799, - "pct_cuda_time": 0.0593360988481595, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.799, - "pct_cuda_time": 0.0593360988481595, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 151.104, - "cuda_time_us": 187.999, - "pct_cuda_time": 0.1963965430263761, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 24.256, - "pct_cuda_time": 0.025339467484655657, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 162.015, - "pct_cuda_time": 0.16925188920376344, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.728, - "pct_cuda_time": 0.0018051863379569993, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.199, - "cuda_time_us": 206.56, - "pct_cuda_time": 0.2157866261391191, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 205.536, - "pct_cuda_time": 0.21471688608699643, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.933, - "cuda_time_us": 44.415, - "pct_cuda_time": 0.046398930092801, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.415, - "pct_cuda_time": 0.046398930092801, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 114.699, - "cuda_time_us": 2181.117, - "pct_cuda_time": 2.2785431770172204, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 48.861, - "cuda_time_us": 1376.894, - "pct_cuda_time": 1.438397128249401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1376.126, - "pct_cuda_time": 1.4375948232103088, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.774, - "cuda_time_us": 180.032, - "pct_cuda_time": 0.1880736729138163, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.032, - "pct_cuda_time": 0.1880736729138163, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.795, - "cuda_time_us": 624.191, - "pct_cuda_time": 0.6520723758540032, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 623.423, - "pct_cuda_time": 0.6512700708149111, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 628.225, - "cuda_time_us": 3031.929, - "pct_cuda_time": 3.1673592641525623, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.131, - "cuda_time_us": 45.024, - "pct_cuda_time": 0.04703513291676849, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.024, - "pct_cuda_time": 0.04703513291676849, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 470.823, - "cuda_time_us": 744.2209999999999, - "pct_cuda_time": 0.7774638782527176, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.573, - "cuda_time_us": 291.71099999999996, - "pct_cuda_time": 0.30474115268042484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0010028812988649995, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 290.751, - "pct_cuda_time": 0.3037382713815599, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.504, - "cuda_time_us": 57.664, - "pct_cuda_time": 0.060239736685157655, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.664, - "pct_cuda_time": 0.060239736685157655, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 244.116, - "cuda_time_us": 188.479, - "pct_cuda_time": 0.19689798367580866, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.616, - "pct_cuda_time": 0.024670879952078992, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 163.295, - "pct_cuda_time": 0.1705890642689168, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.568, - "pct_cuda_time": 0.0016380394548128328, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 47.998, - "cuda_time_us": 206.367, - "pct_cuda_time": 0.21558500521132642, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 205.631, - "pct_cuda_time": 0.21481612954886325, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 21.704, - "cuda_time_us": 44.224, - "pct_cuda_time": 0.046199398501047644, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.224, - "pct_cuda_time": 0.046199398501047644, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.929, - "cuda_time_us": 2198.46, - "pct_cuda_time": 2.2966608544820284, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.355, - "cuda_time_us": 1391.8690000000001, - "pct_cuda_time": 1.4540410318436754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1390.813, - "pct_cuda_time": 1.4529378624149238, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.407, - "cuda_time_us": 180.832, - "pct_cuda_time": 0.1889094073295371, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.832, - "pct_cuda_time": 0.1889094073295371, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.656, - "cuda_time_us": 625.759, - "pct_cuda_time": 0.6537104153088159, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 624.703, - "pct_cuda_time": 0.6526072458800645, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 523.062, - "cuda_time_us": 3027.259, - "pct_cuda_time": 3.162480664500792, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.654, - "cuda_time_us": 45.28, - "pct_cuda_time": 0.04730256792979915, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.28, - "pct_cuda_time": 0.04730256792979915, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 379.254, - "cuda_time_us": 741.854, - "pct_cuda_time": 0.7749911490502036, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.681, - "cuda_time_us": 292.51099999999997, - "pct_cuda_time": 0.3055768870961457, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.0014040338184109996, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 291.167, - "pct_cuda_time": 0.3041728532777347, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 108.44, - "cuda_time_us": 56.576, - "pct_cuda_time": 0.05910313787977731, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.576, - "pct_cuda_time": 0.05910313787977731, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.4, - "cuda_time_us": 186.976, - "pct_cuda_time": 0.1953278476422731, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.36, - "pct_cuda_time": 0.024403444939048325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 162.08, - "pct_cuda_time": 0.1693197926250408, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.0016046100781839994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.823, - "cuda_time_us": 205.791, - "pct_cuda_time": 0.21498327643200743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011365988053803331, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 204.703, - "pct_cuda_time": 0.21384667762662712, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.196, - "cuda_time_us": 45.376, - "pct_cuda_time": 0.047402856059685645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.376, - "pct_cuda_time": 0.047402856059685645, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.702, - "cuda_time_us": 2194.7490000000003, - "pct_cuda_time": 2.2927840914611033, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.576, - "cuda_time_us": 1391.294, - "pct_cuda_time": 1.453440347732376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1390.526, - "pct_cuda_time": 1.452638042693284, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.473, - "cuda_time_us": 180.48, - "pct_cuda_time": 0.18854168418661993, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.48, - "pct_cuda_time": 0.18854168418661993, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.8, - "cuda_time_us": 622.975, - "pct_cuda_time": 0.6508020595421075, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 622.239, - "pct_cuda_time": 0.6500331838796444, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 516.401, - "cuda_time_us": 3024.121, - "pct_cuda_time": 3.1592024962551273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.895, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.046466833514078314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.046466833514078314, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 364.103, - "cuda_time_us": 745.278, - "pct_cuda_time": 0.7785680923494888, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 35.263, - "cuda_time_us": 292.447, - "pct_cuda_time": 0.3055100283428881, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.96, - "pct_cuda_time": 0.0010028812988649995, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 291.487, - "pct_cuda_time": 0.3045071470440231, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.143, - "cuda_time_us": 57.088, - "pct_cuda_time": 0.05963800790583865, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.088, - "pct_cuda_time": 0.05963800790583865, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.563, - "cuda_time_us": 188.48, - "pct_cuda_time": 0.19689902834382828, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.488, - "pct_cuda_time": 0.02453716244556366, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 163.2, - "pct_cuda_time": 0.1704898208070499, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.792, - "pct_cuda_time": 0.0018720450912146662, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.966, - "cuda_time_us": 207.263, - "pct_cuda_time": 0.21652102775693377, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015043219482974994, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 205.823, - "pct_cuda_time": 0.21501670580863627, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.772, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.045564240345099816, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.616, - "pct_cuda_time": 0.045564240345099816, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 105.157, - "cuda_time_us": 2190.7470000000003, - "pct_cuda_time": 2.28860333004646, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.267, - "cuda_time_us": 1385.5330000000001, - "pct_cuda_time": 1.4474220152711663, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1384.477, - "pct_cuda_time": 1.4463188458424148, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 26.699, - "cuda_time_us": 180.864, - "pct_cuda_time": 0.18894283670616593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.864, - "pct_cuda_time": 0.18894283670616593, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.885, - "cuda_time_us": 624.35, - "pct_cuda_time": 0.6522384780691276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 623.294, - "pct_cuda_time": 0.651135308640376, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 522.111, - "cuda_time_us": 3025.3080000000004, - "pct_cuda_time": 3.160442517194453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 20.471, - "cuda_time_us": 45.921, - "pct_cuda_time": 0.04797220013039547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.921, - "pct_cuda_time": 0.04797220013039547, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 372.45, - "cuda_time_us": 741.183, - "pct_cuda_time": 0.7742901768090177, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.711, - "cuda_time_us": 291.488, - "pct_cuda_time": 0.30450819171204274, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.993, - "pct_cuda_time": 0.001037355343513484, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 290.495, - "pct_cuda_time": 0.30347083636852923, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.519, - "cuda_time_us": 57.056, - "pct_cuda_time": 0.059604578529209806, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.056, - "pct_cuda_time": 0.059604578529209806, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 153.718, - "cuda_time_us": 186.815, - "pct_cuda_time": 0.19515965609110927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.552, - "pct_cuda_time": 0.02460402119882132, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 161.535, - "pct_cuda_time": 0.16875044855433097, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.728, - "pct_cuda_time": 0.0018051863379569993, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.855, - "cuda_time_us": 205.824, - "pct_cuda_time": 0.21501775047665594, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 204.768, - "pct_cuda_time": 0.21391458104790445, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.06, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.04690141541025315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.04690141541025315, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.493, - "cuda_time_us": 2193.3080000000004, - "pct_cuda_time": 2.2912787248447866, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.085, - "cuda_time_us": 1388.3490000000002, - "pct_cuda_time": 1.4503638004145039, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1387.613, - "pct_cuda_time": 1.4495949247520405, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.412, - "cuda_time_us": 180.256, - "pct_cuda_time": 0.1883076785502181, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 180.256, - "pct_cuda_time": 0.1883076785502181, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.861, - "cuda_time_us": 624.7030000000001, - "pct_cuda_time": 0.6526072458800646, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 623.647, - "pct_cuda_time": 0.651504076451313, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 528.627, - "cuda_time_us": 3029.5969999999998, - "pct_cuda_time": 3.1649230983307355, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.206, - "cuda_time_us": 44.641, - "pct_cuda_time": 0.046635025065242135, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.641, - "pct_cuda_time": 0.046635025065242135, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 376.161, - "cuda_time_us": 742.113, - "pct_cuda_time": 0.7752617180672933, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.94, - "cuda_time_us": 292.28900000000004, - "pct_cuda_time": 0.30534497079578327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0007699203304828174, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 291.552, - "pct_cuda_time": 0.3045750504653004, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.308, - "cuda_time_us": 57.12, - "pct_cuda_time": 0.05967143728246748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.12, - "pct_cuda_time": 0.05967143728246748, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.53, - "cuda_time_us": 186.84799999999998, - "pct_cuda_time": 0.19519413013575776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.36, - "pct_cuda_time": 0.024403444939048325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 162.08, - "pct_cuda_time": 0.1693197926250408, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.001470892571668666, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 52.219, - "cuda_time_us": 205.856, - "pct_cuda_time": 0.21505117985328473, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 205.088, - "pct_cuda_time": 0.21424887481419275, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.918, - "cuda_time_us": 44.64, - "pct_cuda_time": 0.04663398039722249, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.64, - "pct_cuda_time": 0.04663398039722249, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 103.338, - "cuda_time_us": 2198.203, - "pct_cuda_time": 2.296392374800978, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.988, - "cuda_time_us": 1384.445, - "pct_cuda_time": 1.4462854164657857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.001470892571668666, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1383.037, - "pct_cuda_time": 1.4448145238941172, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.197, - "cuda_time_us": 181.248, - "pct_cuda_time": 0.18934398922571194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 181.248, - "pct_cuda_time": 0.18934398922571194, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.546, - "cuda_time_us": 632.51, - "pct_cuda_time": 0.6607629691094802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 631.774, - "pct_cuda_time": 0.659994093447017, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 503.977, - "cuda_time_us": 3058.166, - "pct_cuda_time": 3.1947682189841466, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.537, - "cuda_time_us": 45.855, - "pct_cuda_time": 0.0479032520410985, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.855, - "pct_cuda_time": 0.0479032520410985, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 359.863, - "cuda_time_us": 752.4140000000001, - "pct_cuda_time": 0.7860228433377187, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.156, - "cuda_time_us": 295.711, - "pct_cuda_time": 0.30891982475902907, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.055, - "pct_cuda_time": 0.0011021247607318485, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.656, - "pct_cuda_time": 0.3078176999982972, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.58, - "cuda_time_us": 57.12, - "pct_cuda_time": 0.05967143728246748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.12, - "pct_cuda_time": 0.05967143728246748, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.482, - "cuda_time_us": 190.239, - "pct_cuda_time": 0.19873659939039445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.024637450575450158, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 165.215, - "pct_cuda_time": 0.1725948268666468, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015043219482974994, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.6, - "cuda_time_us": 209.34400000000002, - "pct_cuda_time": 0.21869498190582762, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.288, - "pct_cuda_time": 0.21759181247707612, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.26, - "cuda_time_us": 44.639, - "pct_cuda_time": 0.04663293572920284, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.639, - "pct_cuda_time": 0.04663293572920284, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.172, - "cuda_time_us": 2215.258, - "pct_cuda_time": 2.314209187876126, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.627, - "cuda_time_us": 1403.2279999999998, - "pct_cuda_time": 1.4659074158788912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007678309944435153, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1402.493, - "pct_cuda_time": 1.4651395848844477, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.666, - "cuda_time_us": 182.463, - "pct_cuda_time": 0.19061326086958796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 182.463, - "pct_cuda_time": 0.19061326086958796, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.206, - "cuda_time_us": 629.567, - "pct_cuda_time": 0.6576885111276471, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011700281820091665, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 628.447, - "pct_cuda_time": 0.656518482945638, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 524.496, - "cuda_time_us": 3055.225, - "pct_cuda_time": 3.1916958503383523, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.143, - "cuda_time_us": 44.64, - "pct_cuda_time": 0.04663398039722249, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.64, - "pct_cuda_time": 0.04663398039722249, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 379.48, - "cuda_time_us": 748.765, - "pct_cuda_time": 0.782210849734012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.48, - "cuda_time_us": 294.591, - "pct_cuda_time": 0.30774979657701995, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.855, - "pct_cuda_time": 0.30698092091455675, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.616, - "cuda_time_us": 57.663, - "pct_cuda_time": 0.060238692017137996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.663, - "pct_cuda_time": 0.060238692017137996, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 177.041, - "cuda_time_us": 188.352, - "pct_cuda_time": 0.1967653108373129, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.024637450575450158, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 163.008, - "pct_cuda_time": 0.17028924454727695, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.76, - "pct_cuda_time": 0.0018386157145858328, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.244, - "cuda_time_us": 208.159, - "pct_cuda_time": 0.21745705030254112, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 207.423, - "pct_cuda_time": 0.21668817464007795, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.355, - "cuda_time_us": 43.999, - "pct_cuda_time": 0.04596434819662617, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 43.999, - "pct_cuda_time": 0.04596434819662617, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 99.589, - "cuda_time_us": 2217.821, - "pct_cuda_time": 2.316886672010492, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.34, - "cuda_time_us": 1406.303, - "pct_cuda_time": 1.4691197700393184, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.057, - "pct_cuda_time": 0.0011042140967711506, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1405.246, - "pct_cuda_time": 1.4680155559425474, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.037, - "cuda_time_us": 181.247, - "pct_cuda_time": 0.18934294455769232, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 181.247, - "pct_cuda_time": 0.18934294455769232, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.352, - "cuda_time_us": 630.271, - "pct_cuda_time": 0.6584239574134814, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.0013706044417821663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 628.959, - "pct_cuda_time": 0.6570533529716992, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 581.909, - "cuda_time_us": 3056.921, - "pct_cuda_time": 3.1934676072996804, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.7, - "cuda_time_us": 45.024, - "pct_cuda_time": 0.04703513291676849, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.024, - "pct_cuda_time": 0.04703513291676849, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 436.273, - "cuda_time_us": 754.782, - "pct_cuda_time": 0.7884966172082524, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.498, - "cuda_time_us": 296.127, - "pct_cuda_time": 0.3093544066552039, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 295.071, - "pct_cuda_time": 0.3082512372264524, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 107.708, - "cuda_time_us": 58.048, - "pct_cuda_time": 0.06064088920470365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 58.048, - "pct_cuda_time": 0.06064088920470365, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 219.938, - "cuda_time_us": 191.008, - "pct_cuda_time": 0.1995399490975061, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.648, - "pct_cuda_time": 0.024704309328707826, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 165.376, - "pct_cuda_time": 0.1727630184178106, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.984, - "pct_cuda_time": 0.002072621350987666, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 46.004, - "cuda_time_us": 209.59900000000002, - "pct_cuda_time": 0.21896137225083864, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.543, - "pct_cuda_time": 0.2178582028220871, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.661, - "cuda_time_us": 44.063, - "pct_cuda_time": 0.046031206949883836, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.063, - "pct_cuda_time": 0.046031206949883836, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.915, - "cuda_time_us": 2213.0519999999997, - "pct_cuda_time": 2.311904650224776, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.149, - "cuda_time_us": 1399.199, - "pct_cuda_time": 1.4616984484277176, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0007699203304828174, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1398.462, - "pct_cuda_time": 1.4609285280972346, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.294, - "cuda_time_us": 182.399, - "pct_cuda_time": 0.19054640211633028, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 182.399, - "pct_cuda_time": 0.19054640211633028, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.82, - "cuda_time_us": 631.454, - "pct_cuda_time": 0.6596597996807286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.055, - "pct_cuda_time": 0.0011021247607318485, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 630.399, - "pct_cuda_time": 0.6585576749199967, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 538.966, - "cuda_time_us": 3058.3289999999997, - "pct_cuda_time": 3.1949384998713493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.298, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.04676769790373782, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.04676769790373782, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 392.572, - "cuda_time_us": 749.981, - "pct_cuda_time": 0.7834811660459077, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.714, - "cuda_time_us": 294.719, - "pct_cuda_time": 0.3078835140835352, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0008023050390919997, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.951, - "pct_cuda_time": 0.3070812090444433, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 120.6, - "cuda_time_us": 56.928, - "pct_cuda_time": 0.059470861022694485, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.928, - "pct_cuda_time": 0.059470861022694485, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 165.419, - "cuda_time_us": 189.311, - "pct_cuda_time": 0.1977671474681583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.744, - "pct_cuda_time": 0.024804597458594327, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 163.967, - "pct_cuda_time": 0.1712910811781223, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.0016714688314416663, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 43.9, - "cuda_time_us": 209.023, - "pct_cuda_time": 0.2183596434715196, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.055, - "pct_cuda_time": 0.0011021247607318485, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 207.968, - "pct_cuda_time": 0.21725751871078774, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.472, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.046466833514078314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.046466833514078314, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.643, - "cuda_time_us": 2219.1, - "pct_cuda_time": 2.3182228024076257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.436, - "cuda_time_us": 1407.965, - "pct_cuda_time": 1.4708560082879785, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1406.941, - "pct_cuda_time": 1.4697862682358558, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.188, - "cuda_time_us": 181.728, - "pct_cuda_time": 0.18984542987514447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 181.728, - "pct_cuda_time": 0.18984542987514447, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.026, - "cuda_time_us": 629.407, - "pct_cuda_time": 0.657521364244503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 628.671, - "pct_cuda_time": 0.6567524885820398, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 528.29, - "cuda_time_us": 3060.728, - "pct_cuda_time": 3.1974446584504927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.865, - "cuda_time_us": 45.344, - "pct_cuda_time": 0.04736942668305682, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.344, - "pct_cuda_time": 0.04736942668305682, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 378.62, - "cuda_time_us": 750.876, - "pct_cuda_time": 0.7844161439234952, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.759, - "cuda_time_us": 295.071, - "pct_cuda_time": 0.3082512372264524, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.0013706044417821663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 293.759, - "pct_cuda_time": 0.30688063278467026, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.9, - "cuda_time_us": 57.28, - "pct_cuda_time": 0.05983858416561164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.28, - "pct_cuda_time": 0.05983858416561164, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.964, - "cuda_time_us": 189.95100000000002, - "pct_cuda_time": 0.198435735000735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.072, - "pct_cuda_time": 0.024102580549388825, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 165.311, - "pct_cuda_time": 0.17269511499653328, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.568, - "pct_cuda_time": 0.0016380394548128328, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 45.478, - "cuda_time_us": 208.574, - "pct_cuda_time": 0.2178905875306963, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0007678309944435153, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 207.839, - "pct_cuda_time": 0.21712275653625276, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.943, - "cuda_time_us": 44.864, - "pct_cuda_time": 0.04686798603362432, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.864, - "pct_cuda_time": 0.04686798603362432, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.315, - "cuda_time_us": 2219.6440000000002, - "pct_cuda_time": 2.3187911018103162, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.585, - "cuda_time_us": 1407.9650000000001, - "pct_cuda_time": 1.4708560082879785, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1407.229, - "pct_cuda_time": 1.4700871326255152, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.994, - "cuda_time_us": 182.176, - "pct_cuda_time": 0.19031344114794807, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 182.176, - "pct_cuda_time": 0.19031344114794807, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.948, - "cuda_time_us": 629.503, - "pct_cuda_time": 0.6576216523743895, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011700281820091665, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 628.383, - "pct_cuda_time": 0.6564516241923803, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 533.968, - "cuda_time_us": 3058.0099999999998, - "pct_cuda_time": 3.1946052507730807, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.662, - "cuda_time_us": 44.831, - "pct_cuda_time": 0.04683351198897583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.831, - "pct_cuda_time": 0.04683351198897583, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 388.551, - "cuda_time_us": 749.888, - "pct_cuda_time": 0.7833840119200802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 32.196, - "cuda_time_us": 294.75300000000004, - "pct_cuda_time": 0.3079190327962034, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0007699203304828174, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.016, - "pct_cuda_time": 0.3071491124657206, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 103.624, - "cuda_time_us": 57.12, - "pct_cuda_time": 0.05967143728246748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.12, - "pct_cuda_time": 0.05967143728246748, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 178.112, - "cuda_time_us": 188.799, - "pct_cuda_time": 0.19723227744209695, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.776, - "pct_cuda_time": 0.024838026835223157, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 163.391, - "pct_cuda_time": 0.17068935239880328, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.0017048982080704995, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 44.062, - "cuda_time_us": 209.216, - "pct_cuda_time": 0.21856126439931228, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.0015043219482974994, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 207.776, - "pct_cuda_time": 0.21705694245101478, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.914, - "cuda_time_us": 44.576, - "pct_cuda_time": 0.046567121643964815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.576, - "pct_cuda_time": 0.046567121643964815, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.4, - "cuda_time_us": 2218.7149999999997, - "pct_cuda_time": 2.3178206052200596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.331, - "cuda_time_us": 1408.1889999999999, - "pct_cuda_time": 1.47109001392438, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.024, - "pct_cuda_time": 0.0010697400521226662, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1407.165, - "pct_cuda_time": 1.4700202738722574, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.254, - "cuda_time_us": 181.312, - "pct_cuda_time": 0.18941084797896962, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 181.312, - "pct_cuda_time": 0.18941084797896962, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.352, - "cuda_time_us": 629.2139999999999, - "pct_cuda_time": 0.6573197433167102, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 628.478, - "pct_cuda_time": 0.6565508676542471, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 520.805, - "cuda_time_us": 3051.834, - "pct_cuda_time": 3.188153381083716, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.067, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.046065680994532315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.046065680994532315, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 372.735, - "cuda_time_us": 750.143, - "pct_cuda_time": 0.7836504022650912, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 40.643, - "cuda_time_us": 295.48699999999997, - "pct_cuda_time": 0.3086858191226272, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.248, - "pct_cuda_time": 0.0013037456885244996, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.239, - "pct_cuda_time": 0.30738207343410273, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.22, - "cuda_time_us": 56.8, - "pct_cuda_time": 0.05933714351617914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 56.8, - "pct_cuda_time": 0.05933714351617914, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 154.453, - "cuda_time_us": 188.704, - "pct_cuda_time": 0.19713303398023013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.936, - "pct_cuda_time": 0.025005173718367326, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 163.168, - "pct_cuda_time": 0.17045639143042113, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.0016714688314416663, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 45.351, - "cuda_time_us": 209.152, - "pct_cuda_time": 0.21849440564605457, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 208.416, - "pct_cuda_time": 0.21772552998359143, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.838, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.046065680994532315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.046065680994532315, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.194, - "cuda_time_us": 2213.499, - "pct_cuda_time": 2.31237161682956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.936, - "cuda_time_us": 1400.7649999999999, - "pct_cuda_time": 1.4633343985464908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.12, - "pct_cuda_time": 0.0011700281820091665, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1399.645, - "pct_cuda_time": 1.4621643703644818, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.57, - "cuda_time_us": 181.727, - "pct_cuda_time": 0.1898443852071248, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 181.727, - "pct_cuda_time": 0.1898443852071248, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.041, - "cuda_time_us": 631.007, - "pct_cuda_time": 0.6591928330759446, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.088, - "pct_cuda_time": 0.0011365988053803331, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 629.919, - "pct_cuda_time": 0.6580562342705643, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 520.159, - "cuda_time_us": 3055.6749999999997, - "pct_cuda_time": 3.192165950947196, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.033, - "cuda_time_us": 45.088, - "pct_cuda_time": 0.047101991670026155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.088, - "pct_cuda_time": 0.047101991670026155, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 361.375, - "cuda_time_us": 751.328, - "pct_cuda_time": 0.7848883338683774, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 39.031, - "cuda_time_us": 295.263, - "pct_cuda_time": 0.30845181348622536, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 294.527, - "pct_cuda_time": 0.3076829378237622, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[4096, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 102.231, - "cuda_time_us": 57.633, - "pct_cuda_time": 0.06020735197654846, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 57.633, - "pct_cuda_time": 0.06020735197654846, - "trace": "_C::rotary_embedding(int64[4096], bfloat16[4096, 4096], bfloat16[4096, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 149.349, - "cuda_time_us": 189.56799999999998, - "pct_cuda_time": 0.19803562714920858, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 23.456, - "pct_cuda_time": 0.024503733068934823, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[4096], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 164.704, - "pct_cuda_time": 0.17206100150860512, - "trace": "_vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.408, - "pct_cuda_time": 0.001470892571668666, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], None, None, bfloat16[4096, 32, 128], int32[3], int32[3], None, None, None, 2048, 2048, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[4096, 32, 128], bfloat16[4096, 8, 128], bfloat16[4096, 8, 128], bfloat16[4096, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.064, - "cuda_time_us": 208.864, - "pct_cuda_time": 0.21819354125639512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 207.808, - "pct_cuda_time": 0.21709037182764357, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[4096, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 23.263, - "cuda_time_us": 44.32, - "pct_cuda_time": 0.04629968663093415, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.32, - "pct_cuda_time": 0.04629968663093415, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.742, - "cuda_time_us": 2214.939, - "pct_cuda_time": 2.3138759387778576, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.223, - "cuda_time_us": 1403.741, - "pct_cuda_time": 1.4664433305729725, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.407, - "pct_cuda_time": 0.0014698479036490152, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 1402.334, - "pct_cuda_time": 1.4649734826693235, - "trace": "mm(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[4096, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[4096, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.555, - "cuda_time_us": 181.951, - "pct_cuda_time": 0.19007839084352662, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 181.951, - "pct_cuda_time": 0.19007839084352662, - "trace": "_C::silu_and_mul(bfloat16[4096, 14336], bfloat16[4096, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 37.745, - "cuda_time_us": 629.247, - "pct_cuda_time": 0.6573542173613588, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0007688756624631663, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 628.511, - "pct_cuda_time": 0.6565853416988956, - "trace": "mm(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[4096, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[4096, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.25, - "cuda_time_us": 45.824, - "pct_cuda_time": 0.04787086733248931, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.824, - "pct_cuda_time": 0.04787086733248931, - "trace": "_C::fused_add_rms_norm(bfloat16[4096, 4096], bfloat16[4096, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 91.451, - "cuda_time_us": 360.288, - "pct_cuda_time": 0.3763813514640344, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.004044954572088832, - "trace": "index_select(bfloat16[4096, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.056, - "pct_cuda_time": 0.0011031694287514998, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 355.36, - "pct_cuda_time": 0.37123322746319404, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 78882.939, - "cuda_time_us": 139.71200000000002, - "pct_cuda_time": 0.1459526583614863, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.003376367039512166, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0022063388575029996, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.002306626987389499, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.002239768234131833, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.002306626987389499, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.002239768234131833, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.002239768234131833, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.448, - "pct_cuda_time": 0.004646683351407832, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 5.248, - "pct_cuda_time": 0.005482417767128665, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 39.52, - "pct_cuda_time": 0.041285280136609155, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 31.744, - "pct_cuda_time": 0.03316194161580266, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.0022731976107606662, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 5.312, - "pct_cuda_time": 0.005549276520386332, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 31.904, - "pct_cuda_time": 0.03332908849894682, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.0033095082862544993, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - }, - "decode_1": { - "metadata": { - "num_running_seqs": 2 - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 6790.903, - "pct_cuda_time": 93.49053716594484, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 6783.927, - "pct_cuda_time": 93.39449839359459, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 211.585, - "pct_cuda_time": 2.9128961650985796, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.353, - "pct_cuda_time": 0.05992786353793565, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 207.232, - "pct_cuda_time": 2.8529683015606437, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 2317.149, - "pct_cuda_time": 31.900250188160822, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 672.4469999999998, - "pct_cuda_time": 9.25759523374551, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 672.4469999999998, - "pct_cuda_time": 9.25759523374551, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 118.623, - "pct_cuda_time": 1.6330859077556954, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 118.623, - "pct_cuda_time": 1.6330859077556954, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 955.396, - "pct_cuda_time": 13.152961431814742, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 74.59400000000002, - "pct_cuda_time": 1.026937526475712, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cuda_time_us": 756.0339999999999, - "pct_cuda_time": 10.408339623716895, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cuda_time_us": 79.007, - "pct_cuda_time": 1.0876914115648249, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 45.761, - "pct_cuda_time": 0.6299928700573107, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 570.683, - "pct_cuda_time": 7.856607614844874, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cuda_time_us": 502.1089999999999, - "pct_cuda_time": 6.912547584003981, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cuda_time_us": 68.574, - "pct_cuda_time": 0.9440600308408913, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 4255.192999999999, - "pct_cuda_time": 58.581352040335176, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 2576.216, - "pct_cuda_time": 35.466832274809654, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2576.216, - "pct_cuda_time": 35.466832274809654, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 287.202, - "pct_cuda_time": 3.953917359021869, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 287.202, - "pct_cuda_time": 3.953917359021869, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 1391.7749999999994, - "pct_cuda_time": 19.160602406503642, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1391.7749999999994, - "pct_cuda_time": 19.160602406503642, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845993100241983, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845993100241983, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 348.28700000000003, - "pct_cuda_time": 4.794876133249942, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04317339307488312, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010132531027778691, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 344.415, - "pct_cuda_time": 4.74157020914728, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 124.543, - "pct_cuda_time": 1.7145867008052196, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 15.744000000000002, - "pct_cuda_time": 0.21674805502900507, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.159, - "pct_cuda_time": 0.05725706052246138, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.768, - "pct_cuda_time": 0.06564117926691414, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 34.176, - "pct_cuda_time": 0.4705018755507671, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.648, - "pct_cuda_time": 0.38063073078264303, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 2.016, - "pct_cuda_time": 0.02775432411956772, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.896, - "pct_cuda_time": 0.06740335857609303, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 28.384, - "pct_cuda_time": 0.39076326181042176, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.752, - "pct_cuda_time": 0.03788685514734641, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 17424.144, - "cuda_time_us": 6790.903, - "pct_cuda_time": 93.49053716594484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 53.645, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[2]) <- embedding(bfloat16[128256, 4096], int64[2], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 836.445, - "cuda_time_us": 214.849, - "pct_cuda_time": 2.9578317374826413, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 50.787, - "cuda_time_us": 4.353, - "pct_cuda_time": 0.05992786353793565, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.353, - "pct_cuda_time": 0.05992786353793565, - "trace": "_C::rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 609.147, - "cuda_time_us": 74.592, - "pct_cuda_time": 1.0269099924240057, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 92.381, - "cuda_time_us": 23.36, - "pct_cuda_time": 0.32159772392514974, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 23.36, - "pct_cuda_time": 0.32159772392514974, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 179.458, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 225.917, - "cuda_time_us": 29.696, - "pct_cuda_time": 0.40882559972950544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.335, - "pct_cuda_time": 0.03214600536666202, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.616, - "pct_cuda_time": 0.32512208254350755, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.433, - "pct_cuda_time": 0.03349517390025211, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.26, - "cuda_time_us": 17.887999999999998, - "pct_cuda_time": 0.24626455845775164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.776, - "pct_cuda_time": 0.21718859985629976, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 23.992, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 127.475, - "cuda_time_us": 132.736, - "pct_cuda_time": 1.827379943618522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 48.221, - "cuda_time_us": 80.736, - "pct_cuda_time": 1.111494599264593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.736, - "pct_cuda_time": 1.111494599264593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 28.776, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.11938764819687069, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.672, - "pct_cuda_time": 0.11938764819687069, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 34.667, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.5964976961570586, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.5964976961570586, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 555.878, - "cuda_time_us": 212.576, - "pct_cuda_time": 2.926539287718863, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.721, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 397.604, - "cuda_time_us": 73.50399999999999, - "pct_cuda_time": 1.011931468295985, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.597, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.2947244894601715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.408, - "pct_cuda_time": 0.2947244894601715, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 113.999, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.05462755858454599, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.05462755858454599, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 181.067, - "cuda_time_us": 30.176, - "pct_cuda_time": 0.41543377213892635, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.84, - "pct_cuda_time": 0.32820589633457065, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436249652898861, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.021146151710146836, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.268, - "cuda_time_us": 17.951999999999998, - "pct_cuda_time": 0.2471456481123411, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.872, - "pct_cuda_time": 0.21851023433818395, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.028635413774157174, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.946, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.456, - "cuda_time_us": 132.416, - "pct_cuda_time": 1.822974495345575, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.564, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1053269716824667, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.288, - "pct_cuda_time": 1.1053269716824667, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.592, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.12379309646981794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.12379309646981794, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 33.708, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.5938544271932903, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.5938544271932903, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 520.061, - "cuda_time_us": 212.70299999999997, - "pct_cuda_time": 2.9282877000021883, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.722, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 365.054, - "cuda_time_us": 72.15899999999999, - "pct_cuda_time": 0.9934148185237534, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.41, - "cuda_time_us": 20.863, - "pct_cuda_time": 0.2872214603703082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.863, - "pct_cuda_time": 0.2872214603703082, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.965, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.879, - "cuda_time_us": 29.695999999999998, - "pct_cuda_time": 0.40882559972950544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.488, - "pct_cuda_time": 0.32335990323432867, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03304086204710442, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.02070560688285211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 37.099, - "cuda_time_us": 17.951999999999998, - "pct_cuda_time": 0.2471456481123411, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.808, - "pct_cuda_time": 0.21762914468359448, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.029516503428746628, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.264, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.406, - "cuda_time_us": 134.016, - "pct_cuda_time": 1.845001736710311, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.011, - "cuda_time_us": 81.6, - "pct_cuda_time": 1.1233893096015504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.6, - "pct_cuda_time": 1.1233893096015504, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.375, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.12819854474276518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.12819854474276518, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.357, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.5934138823659956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.5934138823659956, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 550.712, - "cuda_time_us": 212.00100000000003, - "pct_cuda_time": 2.9186232478534113, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.207, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 393.303, - "cuda_time_us": 72.60900000000001, - "pct_cuda_time": 0.9996099801575857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.058, - "cuda_time_us": 21.409, - "pct_cuda_time": 0.2947382564860245, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.409, - "pct_cuda_time": 0.2947382564860245, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 113.469, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 178.405, - "cuda_time_us": 30.08, - "pct_cuda_time": 0.4141121376570422, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.32468153771621283, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.656, - "pct_cuda_time": 0.036565220665462236, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.021146151710146836, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 37.541, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.24053747570292025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.21146151710146835, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.305, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 104.201, - "cuda_time_us": 132.544, - "pct_cuda_time": 1.824736674654754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.74, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1026837027186984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1026837027186984, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.772, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.11850655854228123, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.11850655854228123, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.245, - "cuda_time_us": 43.84, - "pct_cuda_time": 0.6035464133937742, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.84, - "pct_cuda_time": 0.6035464133937742, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 504.616, - "cuda_time_us": 210.56, - "pct_cuda_time": 2.898784963599295, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.434, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.04978156548430401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.04978156548430401, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 361.43, - "cuda_time_us": 71.616, - "pct_cuda_time": 0.9859393234855962, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.752, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2837108687778034, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2837108687778034, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.635, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.303, - "cuda_time_us": 29.823999999999998, - "pct_cuda_time": 0.4105877790386844, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.68, - "pct_cuda_time": 0.32600317219809705, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436249652898861, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.01850288274637848, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.699, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2414185653575097, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.21234260675605782, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.202, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 95.054, - "cuda_time_us": 132.16, - "pct_cuda_time": 1.8194501367272173, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.276, - "cuda_time_us": 79.616, - "pct_cuda_time": 1.0960755303092775, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.616, - "pct_cuda_time": 1.0960755303092775, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.755, - "cuda_time_us": 9.408, - "pct_cuda_time": 0.12952017922464937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.408, - "pct_cuda_time": 0.12952017922464937, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.738, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.5938544271932903, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.5938544271932903, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 578.196, - "cuda_time_us": 209.88600000000002, - "pct_cuda_time": 2.8895059881744003, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.185, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 421.16, - "cuda_time_us": 71.583, - "pct_cuda_time": 0.9854850116324485, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.429, - "cuda_time_us": 20.703, - "pct_cuda_time": 0.28501873623383456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.703, - "pct_cuda_time": 0.28501873623383456, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 119.704, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.04978156548430401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.04978156548430401, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 172.2, - "cuda_time_us": 29.503999999999998, - "pct_cuda_time": 0.40618233076573707, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.488, - "pct_cuda_time": 0.32335990323432867, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03304086204710442, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 40.925, - "cuda_time_us": 17.759999999999998, - "pct_cuda_time": 0.24450237914857273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.68, - "pct_cuda_time": 0.21586696537441558, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.028635413774157174, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.432, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.0444950275567673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.0444950275567673, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 103.148, - "cuda_time_us": 131.711, - "pct_cuda_time": 1.8132687421192382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.806, - "cuda_time_us": 79.615, - "pct_cuda_time": 1.0960617632834246, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.615, - "pct_cuda_time": 1.0960617632834246, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.939, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894710336957597, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.11894710336957597, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.069, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.5982598754662376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.5982598754662376, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 495.7, - "cuda_time_us": 211.80700000000002, - "pct_cuda_time": 2.9159524448379366, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.629, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 354.688, - "cuda_time_us": 71.935, - "pct_cuda_time": 0.9903310047326905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.591, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2837108687778034, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2837108687778034, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.649, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05110319996618818, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05110319996618818, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.115, - "cuda_time_us": 29.886999999999997, - "pct_cuda_time": 0.41145510166742083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.648, - "pct_cuda_time": 0.3255626273708023, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.471, - "pct_cuda_time": 0.020251295029704425, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.538, - "cuda_time_us": 17.728, - "pct_cuda_time": 0.24406183432127806, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.21498587571982614, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.376, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.999, - "cuda_time_us": 133.344, - "pct_cuda_time": 1.835750295337122, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.301, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.1035647923732879, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.1035647923732879, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.871, - "cuda_time_us": 9.024, - "pct_cuda_time": 0.12423364129711265, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.024, - "pct_cuda_time": 0.12423364129711265, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.517, - "cuda_time_us": 44.16, - "pct_cuda_time": 0.6079518616667214, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.16, - "pct_cuda_time": 0.6079518616667214, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 505.091, - "cuda_time_us": 210.975, - "pct_cuda_time": 2.9044982793282736, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.45, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 357.476, - "cuda_time_us": 71.839, - "pct_cuda_time": 0.9890093702508063, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.122, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.28327032395050866, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.28327032395050866, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.898, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05110319996618818, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.05110319996618818, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.164, - "cuda_time_us": 30.016, - "pct_cuda_time": 0.4132310480024527, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.744, - "pct_cuda_time": 0.3268842618526865, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.463, - "pct_cuda_time": 0.03390818467584092, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.020719373908705065, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 36.721, - "cuda_time_us": 17.535, - "pct_cuda_time": 0.24140479833165673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.392, - "pct_cuda_time": 0.2119020619287631, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.02950273640289366, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.252, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04317339307488312, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04317339307488312, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 100.037, - "cuda_time_us": 132.672, - "pct_cuda_time": 1.8264988539639329, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.715, - "cuda_time_us": 79.84, - "pct_cuda_time": 1.0991593441003407, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.84, - "pct_cuda_time": 1.0991593441003407, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.002, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.12819854474276518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.12819854474276518, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.392, - "cuda_time_us": 43.52, - "pct_cuda_time": 0.599140965120827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.52, - "pct_cuda_time": 0.599140965120827, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 487.663, - "cuda_time_us": 211.58300000000003, - "pct_cuda_time": 2.912868631046874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.916, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 346.862, - "cuda_time_us": 72.28800000000001, - "pct_cuda_time": 0.9951907648587854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.15, - "cuda_time_us": 21.088, - "pct_cuda_time": 0.29031904118722424, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.088, - "pct_cuda_time": 0.29031904118722424, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.805, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.04934102065700929, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.04934102065700929, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.41, - "cuda_time_us": 29.793, - "pct_cuda_time": 0.41016100123724264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.585, - "pct_cuda_time": 0.32469530474206587, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.033481406874399156, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.943, - "cuda_time_us": 17.823, - "pct_cuda_time": 0.24536970177730927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.712, - "pct_cuda_time": 0.2163075102017103, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.02906219157559894, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.211, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.043159626049030154, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.043159626049030154, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.023, - "cuda_time_us": 132.8, - "pct_cuda_time": 1.8282610332731117, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.212, - "cuda_time_us": 79.936, - "pct_cuda_time": 1.100480978582225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.936, - "pct_cuda_time": 1.100480978582225, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.201, - "cuda_time_us": 9.472, - "pct_cuda_time": 0.1304012688792388, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.472, - "pct_cuda_time": 0.1304012688792388, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.994, - "cuda_time_us": 43.392, - "pct_cuda_time": 0.597378785811648, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.392, - "pct_cuda_time": 0.597378785811648, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 521.056, - "cuda_time_us": 210.624, - "pct_cuda_time": 2.8996660532538847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.757, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.046697751693240926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.046697751693240926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 380.707, - "cuda_time_us": 71.968, - "pct_cuda_time": 0.9907853165858382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.706, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.28547304808698226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.28547304808698226, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.37, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242483444807236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242483444807236, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 158.833, - "cuda_time_us": 29.76, - "pct_cuda_time": 0.40970668938409494, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.713, - "pct_cuda_time": 0.3264574840512447, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.431, - "pct_cuda_time": 0.0334676398485462, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.041, - "cuda_time_us": 17.664, - "pct_cuda_time": 0.2431807446666886, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.584, - "pct_cuda_time": 0.2145453308925314, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.028635413774157174, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.421, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.044935572384062025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.044935572384062025, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.609, - "cuda_time_us": 132.0, - "pct_cuda_time": 1.8172474125907434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.98, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.1035647923732879, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.1035647923732879, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.783, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511473095170209, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511473095170209, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.529, - "cuda_time_us": 42.752, - "pct_cuda_time": 0.5885678892657535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.752, - "pct_cuda_time": 0.5885678892657535, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 509.984, - "cuda_time_us": 214.525, - "pct_cuda_time": 2.9533712211062824, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.834, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 365.993, - "cuda_time_us": 74.014, - "pct_cuda_time": 1.0189526514809946, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.539, - "cuda_time_us": 21.792, - "pct_cuda_time": 0.30001102738770824, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.792, - "pct_cuda_time": 0.30001102738770824, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 107.378, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.05330592410266181, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.05330592410266181, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 162.858, - "cuda_time_us": 29.886, - "pct_cuda_time": 0.4114413346415679, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03127868273792552, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.648, - "pct_cuda_time": 0.3255626273708023, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.463, - "pct_cuda_time": 0.03390818467584092, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.020691839856999145, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.542, - "cuda_time_us": 18.464, - "pct_cuda_time": 0.2541943653490567, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.352, - "pct_cuda_time": 0.22511840674760483, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.287, - "cuda_time_us": 3.295, - "pct_cuda_time": 0.04536235018550379, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.295, - "pct_cuda_time": 0.04536235018550379, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.793, - "cuda_time_us": 133.888, - "pct_cuda_time": 1.8432395574011324, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.551, - "cuda_time_us": 81.664, - "pct_cuda_time": 1.12427039925614, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.664, - "pct_cuda_time": 1.12427039925614, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.672, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.12379309646981794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.12379309646981794, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.961, - "cuda_time_us": 43.232, - "pct_cuda_time": 0.5951760616751745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.232, - "pct_cuda_time": 0.5951760616751745, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 557.855, - "cuda_time_us": 211.649, - "pct_cuda_time": 2.913777254753169, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.383, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.045376117211356745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.045376117211356745, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 402.451, - "cuda_time_us": 71.87299999999999, - "pct_cuda_time": 0.9894774491298068, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.094, - "cuda_time_us": 20.832, - "pct_cuda_time": 0.2867946825688665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.832, - "pct_cuda_time": 0.2867946825688665, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.191, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 174.927, - "cuda_time_us": 29.825999999999997, - "pct_cuda_time": 0.4106153130903903, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.032159772392514975, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.521, - "pct_cuda_time": 0.3238142150874764, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.020719373908705065, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.44, - "cuda_time_us": 17.535, - "pct_cuda_time": 0.24140479833165673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.423, - "pct_cuda_time": 0.21232883973020486, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.728, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.533, - "cuda_time_us": 133.152, - "pct_cuda_time": 1.8331070263733535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.686, - "cuda_time_us": 80.928, - "pct_cuda_time": 1.1141378682283614, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.928, - "pct_cuda_time": 1.1141378682283614, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.953, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114982750604958, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114982750604958, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.707, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.5978193306389429, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.5978193306389429, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 588.617, - "cuda_time_us": 213.12, - "pct_cuda_time": 2.9340285497828735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.298, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.046697751693240926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.046697751693240926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 439.967, - "cuda_time_us": 72.96000000000001, - "pct_cuda_time": 1.0044422062319747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 31.057, - "cuda_time_us": 21.376, - "pct_cuda_time": 0.29428394463287677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.376, - "pct_cuda_time": 0.29428394463287677, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 110.271, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 229.046, - "cuda_time_us": 30.048, - "pct_cuda_time": 0.41367159282974747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.872, - "pct_cuda_time": 0.32864644116186537, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.033481406874399156, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.44, - "pct_cuda_time": 0.019824517228262655, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.27, - "cuda_time_us": 17.856, - "pct_cuda_time": 0.24582401363045697, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.712, - "pct_cuda_time": 0.2163075102017103, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.029516503428746628, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.791, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.102, - "cuda_time_us": 133.568, - "pct_cuda_time": 1.8388341091281852, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.805, - "cuda_time_us": 81.696, - "pct_cuda_time": 1.1247109440834349, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.696, - "pct_cuda_time": 1.1247109440834349, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.051, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.11982819302416539, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.11982819302416539, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.752, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.594294972020585, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.594294972020585, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 487.544, - "cuda_time_us": 210.429, - "pct_cuda_time": 2.8969814832125573, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.173, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.044935572384062025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.044935572384062025, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 349.35, - "cuda_time_us": 71.774, - "pct_cuda_time": 0.9881145135703638, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.182, - "cuda_time_us": 20.543, - "pct_cuda_time": 0.28281601209736096, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.543, - "pct_cuda_time": 0.28281601209736096, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 98.794, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.04978156548430401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.04978156548430401, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 159.801, - "cuda_time_us": 29.727999999999998, - "pct_cuda_time": 0.40926614455680016, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.033481406874399156, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.584, - "pct_cuda_time": 0.32468153771621283, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03304086204710442, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.812, - "cuda_time_us": 17.887, - "pct_cuda_time": 0.24625079143189874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.743, - "pct_cuda_time": 0.2167342880031521, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.029516503428746628, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.206, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.044935572384062025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.044935572384062025, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.235, - "cuda_time_us": 132.127, - "pct_cuda_time": 1.8189958248740696, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.467, - "cuda_time_us": 80.447, - "pct_cuda_time": 1.1075159287930876, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.447, - "pct_cuda_time": 1.1075159287930876, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.601, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12291200681522849, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12291200681522849, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.047, - "cuda_time_us": 42.752, - "pct_cuda_time": 0.5885678892657535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.752, - "pct_cuda_time": 0.5885678892657535, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 506.117, - "cuda_time_us": 211.327, - "pct_cuda_time": 2.9093442724285157, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.421, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.046697751693240926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.392, - "pct_cuda_time": 0.046697751693240926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 364.2, - "cuda_time_us": 72.0, - "pct_cuda_time": 0.991225861413133, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.332, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.28591359291427704, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.28591359291427704, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 170.36, - "cuda_time_us": 29.727999999999998, - "pct_cuda_time": 0.40926614455680016, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381379106308, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.52, - "pct_cuda_time": 0.3238004480616234, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.02070560688285211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.655, - "cuda_time_us": 17.856, - "pct_cuda_time": 0.24582401363045697, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.712, - "pct_cuda_time": 0.2163075102017103, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.029516503428746628, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.382, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.254, - "cuda_time_us": 132.767, - "pct_cuda_time": 1.827806721419964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.856, - "cuda_time_us": 80.703, - "pct_cuda_time": 1.1110402874114453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.703, - "pct_cuda_time": 1.1110402874114453, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.986, - "cuda_time_us": 8.769, - "pct_cuda_time": 0.1207230497046078, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.769, - "pct_cuda_time": 0.1207230497046078, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.328, - "cuda_time_us": 43.295, - "pct_cuda_time": 0.5960433843039109, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.295, - "pct_cuda_time": 0.5960433843039109, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 478.095, - "cuda_time_us": 212.127, - "pct_cuda_time": 2.920357893110884, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.962, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 342.177, - "cuda_time_us": 72.928, - "pct_cuda_time": 1.0040016614046798, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.137, - "cuda_time_us": 21.376, - "pct_cuda_time": 0.29428394463287677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.376, - "pct_cuda_time": 0.29428394463287677, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 95.631, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 155.68, - "cuda_time_us": 29.951999999999998, - "pct_cuda_time": 0.41234995834786325, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.68, - "pct_cuda_time": 0.32600317219809705, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.02070560688285211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.748, - "cuda_time_us": 17.92, - "pct_cuda_time": 0.2467051032850464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.808, - "pct_cuda_time": 0.21762914468359448, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.217, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 92.195, - "cuda_time_us": 132.671, - "pct_cuda_time": 1.82648508693808, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.146, - "cuda_time_us": 81.248, - "pct_cuda_time": 1.1185433165013086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.248, - "pct_cuda_time": 1.1185433165013086, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.557, - "cuda_time_us": 8.799, - "pct_cuda_time": 0.1211360604801966, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.799, - "pct_cuda_time": 0.1211360604801966, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.434, - "cuda_time_us": 42.624, - "pct_cuda_time": 0.5868057099565747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.624, - "pct_cuda_time": 0.5868057099565747, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 491.031, - "cuda_time_us": 211.553, - "pct_cuda_time": 2.9124556202712846, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.802, - "cuda_time_us": 3.361, - "pct_cuda_time": 0.04627097389179916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.361, - "pct_cuda_time": 0.04627097389179916, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 347.187, - "cuda_time_us": 71.80699999999999, - "pct_cuda_time": 0.9885688254235114, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.162, - "cuda_time_us": 20.863, - "pct_cuda_time": 0.2872214603703082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.863, - "pct_cuda_time": 0.2872214603703082, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 93.801, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.788, - "cuda_time_us": 29.855999999999998, - "pct_cuda_time": 0.4110283238659791, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.032159772392514975, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.712, - "pct_cuda_time": 0.32644371702539177, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.01850288274637848, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.933, - "cuda_time_us": 17.439999999999998, - "pct_cuda_time": 0.24009693087562547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.21146151710146835, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.028635413774157174, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.134, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 97.847, - "cuda_time_us": 133.21699999999998, - "pct_cuda_time": 1.8340018830537956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.911, - "cuda_time_us": 80.832, - "pct_cuda_time": 1.112816233746477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.832, - "pct_cuda_time": 1.112816233746477, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.818, - "cuda_time_us": 9.152, - "pct_cuda_time": 0.12599582060629155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.152, - "pct_cuda_time": 0.12599582060629155, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.525, - "cuda_time_us": 43.233, - "pct_cuda_time": 0.5951898287010273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.233, - "pct_cuda_time": 0.5951898287010273, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 487.914, - "cuda_time_us": 210.24, - "pct_cuda_time": 2.894379515326348, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.065, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 348.886, - "cuda_time_us": 72.12899999999999, - "pct_cuda_time": 0.9930018077481647, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.738, - "cuda_time_us": 20.833, - "pct_cuda_time": 0.2868084495947194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.833, - "pct_cuda_time": 0.2868084495947194, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.705, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.04934102065700929, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.04934102065700929, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 152.739, - "cuda_time_us": 29.983999999999998, - "pct_cuda_time": 0.412790503175158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436249652898861, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.744, - "pct_cuda_time": 0.3268842618526865, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.033481406874399156, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 32.912, - "cuda_time_us": 17.728, - "pct_cuda_time": 0.24406183432127806, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.21498587571982614, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.326, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04317339307488312, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04317339307488312, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.198, - "cuda_time_us": 131.615, - "pct_cuda_time": 1.8119471076373541, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.323, - "cuda_time_us": 80.031, - "pct_cuda_time": 1.101788846038256, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.031, - "pct_cuda_time": 1.101788846038256, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.301, - "cuda_time_us": 8.609, - "pct_cuda_time": 0.11852032556813416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.609, - "pct_cuda_time": 0.11852032556813416, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.378, - "cuda_time_us": 42.975, - "pct_cuda_time": 0.5916379360309637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.975, - "pct_cuda_time": 0.5916379360309637, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 502.644, - "cuda_time_us": 212.83100000000002, - "pct_cuda_time": 2.930049879311368, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.987, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.045376117211356745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.045376117211356745, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 361.305, - "cuda_time_us": 72.351, - "pct_cuda_time": 0.9960580874875219, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.361, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.28591359291427704, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.28591359291427704, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.584, - "cuda_time_us": 3.679, - "pct_cuda_time": 0.0506488881130405, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.679, - "pct_cuda_time": 0.0506488881130405, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.772, - "cuda_time_us": 29.727999999999998, - "pct_cuda_time": 0.40926614455680016, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03127868273792552, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.392, - "pct_cuda_time": 0.3220382687524445, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03524358618357806, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.02070560688285211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.581, - "cuda_time_us": 18.176, - "pct_cuda_time": 0.25022946190340417, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.84, - "pct_cuda_time": 0.2180696895108892, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.032159772392514975, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.84, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04317339307488312, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04317339307488312, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 96.885, - "cuda_time_us": 134.048, - "pct_cuda_time": 1.845442281537606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 35.041, - "cuda_time_us": 81.791, - "pct_cuda_time": 1.126018811539466, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.791, - "pct_cuda_time": 1.126018811539466, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.884, - "cuda_time_us": 8.929, - "pct_cuda_time": 0.12292577384108144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.929, - "pct_cuda_time": 0.12292577384108144, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.888, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.5964976961570586, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.328, - "pct_cuda_time": 0.5964976961570586, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 604.279, - "cuda_time_us": 212.47699999999998, - "pct_cuda_time": 2.9251763521594194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.446, - "cuda_time_us": 3.231, - "pct_cuda_time": 0.044481260530914335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.231, - "pct_cuda_time": 0.044481260530914335, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 445.409, - "cuda_time_us": 73.05499999999999, - "pct_cuda_time": 1.0057500736880058, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.484, - "cuda_time_us": 21.248, - "pct_cuda_time": 0.2925217653236979, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.248, - "pct_cuda_time": 0.2925217653236979, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 101.847, - "cuda_time_us": 3.839, - "pct_cuda_time": 0.052851612249514124, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.839, - "pct_cuda_time": 0.052851612249514124, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 244.113, - "cuda_time_us": 30.144, - "pct_cuda_time": 0.41499322731163163, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.368, - "pct_cuda_time": 0.0326003172198097, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.648, - "pct_cuda_time": 0.3255626273708023, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.022908331019325736, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 41.021, - "cuda_time_us": 17.823999999999998, - "pct_cuda_time": 0.2453834688031622, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.712, - "pct_cuda_time": 0.2163075102017103, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 19.158, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 110.175, - "cuda_time_us": 133.023, - "pct_cuda_time": 1.8313310800383218, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 41.54, - "cuda_time_us": 80.543, - "pct_cuda_time": 1.1088375632749718, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.543, - "pct_cuda_time": 1.1088375632749718, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 24.043, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.1215903723333443, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.1215903723333443, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 32.039, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6009031444300059, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6009031444300059, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 542.76, - "cuda_time_us": 212.162, - "pct_cuda_time": 2.9208397390157375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.762, - "cuda_time_us": 3.297, - "pct_cuda_time": 0.04538988423720971, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.297, - "pct_cuda_time": 0.04538988423720971, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 391.714, - "cuda_time_us": 73.15299999999999, - "pct_cuda_time": 1.0070992422215959, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.666, - "cuda_time_us": 22.113, - "pct_cuda_time": 0.3044302426865084, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 22.113, - "pct_cuda_time": 0.3044302426865084, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.597, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242483444807236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242483444807236, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 181.47, - "cuda_time_us": 29.439999999999998, - "pct_cuda_time": 0.4053012411111476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.328, - "pct_cuda_time": 0.321157179097855, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.01850288274637848, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.222, - "cuda_time_us": 17.791999999999998, - "pct_cuda_time": 0.24494292397586748, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.488, - "pct_cuda_time": 0.21322369641064723, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 17.654, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.463, - "cuda_time_us": 132.544, - "pct_cuda_time": 1.824736674654754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.207, - "cuda_time_us": 79.36, - "pct_cuda_time": 1.0925511716909198, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.36, - "pct_cuda_time": 1.0925511716909198, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.459, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.12379309646981794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.12379309646981794, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.295, - "cuda_time_us": 44.192, - "pct_cuda_time": 0.6083924064940162, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.192, - "pct_cuda_time": 0.6083924064940162, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 493.523, - "cuda_time_us": 211.392, - "pct_cuda_time": 2.910239129108958, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.062, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 355.093, - "cuda_time_us": 71.264, - "pct_cuda_time": 0.9810933303853541, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.035, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2837108687778034, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2837108687778034, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 100.887, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 162.483, - "cuda_time_us": 29.439999999999998, - "pct_cuda_time": 0.4053012411111476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.273, - "pct_cuda_time": 0.03129244976377849, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.423, - "pct_cuda_time": 0.32246504655388625, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.033481406874399156, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.489, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2414185653575097, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.456, - "pct_cuda_time": 0.21278315158335254, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.028635413774157174, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.247, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.225, - "cuda_time_us": 133.344, - "pct_cuda_time": 1.835750295337122, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.035, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1026837027186984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1026837027186984, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.627, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511473095170209, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511473095170209, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.932, - "cuda_time_us": 44.16, - "pct_cuda_time": 0.6079518616667214, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.16, - "pct_cuda_time": 0.6079518616667214, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 524.518, - "cuda_time_us": 213.21599999999995, - "pct_cuda_time": 2.935350184264757, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.81, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 373.915, - "cuda_time_us": 72.12799999999999, - "pct_cuda_time": 0.9929880407223115, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.914, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.28591359291427704, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.28591359291427704, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.772, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 161.114, - "cuda_time_us": 29.823999999999998, - "pct_cuda_time": 0.4105877790386844, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.337, - "pct_cuda_time": 0.032173539418367945, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.52, - "pct_cuda_time": 0.3238004480616234, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.020691839856999145, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 48.138, - "cuda_time_us": 17.887999999999998, - "pct_cuda_time": 0.24626455845775164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.744, - "pct_cuda_time": 0.21674805502900504, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.029516503428746628, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.418, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.0444950275567673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.0444950275567673, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 102.352, - "cuda_time_us": 134.39999999999998, - "pct_cuda_time": 1.8502882746378477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.332, - "cuda_time_us": 81.216, - "pct_cuda_time": 1.1181027716740137, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.216, - "pct_cuda_time": 1.1181027716740137, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.095, - "cuda_time_us": 9.44, - "pct_cuda_time": 0.1299607240519441, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.44, - "pct_cuda_time": 0.1299607240519441, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 31.135, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.60222477891189, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.60222477891189, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 507.386, - "cuda_time_us": 212.44899999999998, - "pct_cuda_time": 2.9247908754355367, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.904, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 355.899, - "cuda_time_us": 72.38499999999999, - "pct_cuda_time": 0.9965261663665225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.244, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841514136050981, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841514136050981, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 99.228, - "cuda_time_us": 3.905, - "pct_cuda_time": 0.0537602359558095, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.905, - "pct_cuda_time": 0.0537602359558095, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 162.097, - "cuda_time_us": 30.208, - "pct_cuda_time": 0.41587431696622107, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.872, - "pct_cuda_time": 0.32864644116186537, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03436249652898861, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.021146151710146836, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 34.746, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.24274019983939382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.21366424123794195, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.806, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 103.059, - "cuda_time_us": 133.504, - "pct_cuda_time": 1.8379530194735954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.543, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1088513303008247, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1088513303008247, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 27.6, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114982750604958, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114982750604958, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.403, - "cuda_time_us": 44.16, - "pct_cuda_time": 0.6079518616667214, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.16, - "pct_cuda_time": 0.6079518616667214, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 485.586, - "cuda_time_us": 211.328, - "pct_cuda_time": 2.9093580394543688, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.503, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 347.487, - "cuda_time_us": 71.93599999999999, - "pct_cuda_time": 0.9903447717585433, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 34.541, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.2876757722234559, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.2876757722234559, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 93.815, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.04978156548430401, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.04978156548430401, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 156.245, - "cuda_time_us": 29.503999999999998, - "pct_cuda_time": 0.40618233076573707, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.368, - "pct_cuda_time": 0.0326003172198097, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.424, - "pct_cuda_time": 0.32247881357973923, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03304086204710442, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.588, - "cuda_time_us": 17.919999999999998, - "pct_cuda_time": 0.24670510328504638, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.616, - "pct_cuda_time": 0.21498587571982614, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.005, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 92.747, - "cuda_time_us": 132.768, - "pct_cuda_time": 1.827820488445817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.983, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1026837027186984, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.1026837027186984, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.488, - "cuda_time_us": 9.024, - "pct_cuda_time": 0.12423364129711265, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.024, - "pct_cuda_time": 0.12423364129711265, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.224, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6009031444300059, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6009031444300059, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 491.513, - "cuda_time_us": 211.904, - "pct_cuda_time": 2.917287846345674, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.893, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019386175125106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.488, - "pct_cuda_time": 0.048019386175125106, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 347.42, - "cuda_time_us": 71.744, - "pct_cuda_time": 0.9877015027947751, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.869, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.28503250325968754, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.28503250325968754, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 97.194, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.04934102065700929, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.04934102065700929, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.529, - "cuda_time_us": 29.887999999999998, - "pct_cuda_time": 0.4114688686932738, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.591, - "pct_cuda_time": 0.035670363985019826, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.52, - "pct_cuda_time": 0.3238004480616234, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.313, - "pct_cuda_time": 0.018076104944936715, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.618, - "cuda_time_us": 17.567999999999998, - "pct_cuda_time": 0.24185911018480438, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.488, - "pct_cuda_time": 0.21322369641064723, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.028635413774157174, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.451, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.045376117211356745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.045376117211356745, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.407, - "cuda_time_us": 133.376, - "pct_cuda_time": 1.836190840164417, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.315, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.113256778573772, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.113256778573772, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.357, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114982750604958, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12114982750604958, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.422, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6017842340845954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6017842340845954, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 554.77, - "cuda_time_us": 212.00100000000003, - "pct_cuda_time": 2.9186232478534113, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.609, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.04757884134783038, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 407.295, - "cuda_time_us": 72.257, - "pct_cuda_time": 0.9947639870573437, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 36.099, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.2845919584323928, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.2845919584323928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 94.355, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242483444807236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242483444807236, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 209.948, - "cuda_time_us": 29.953, - "pct_cuda_time": 0.41236372537371624, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.273, - "pct_cuda_time": 0.03129244976377849, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.712, - "pct_cuda_time": 0.32644371702539177, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.02070560688285211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 37.039, - "cuda_time_us": 17.823999999999998, - "pct_cuda_time": 0.2453834688031622, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.584, - "pct_cuda_time": 0.2145453308925314, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381379106308, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.587, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.306, - "cuda_time_us": 133.12, - "pct_cuda_time": 1.8326664815460592, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 36.522, - "cuda_time_us": 80.672, - "pct_cuda_time": 1.1106135096100036, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.672, - "pct_cuda_time": 1.1106135096100036, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.249, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511473095170209, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511473095170209, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.892, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.5969382409843533, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.5969382409843533, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 518.157, - "cuda_time_us": 212.22500000000002, - "pct_cuda_time": 2.9217070616444745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.705, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 358.285, - "cuda_time_us": 73.313, - "pct_cuda_time": 1.0093019663580696, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 33.404, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.29340285497828733, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.312, - "pct_cuda_time": 0.29340285497828733, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 93.308, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 157.939, - "cuda_time_us": 29.857, - "pct_cuda_time": 0.4110420908918321, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.337, - "pct_cuda_time": 0.032173539418367945, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.552, - "pct_cuda_time": 0.3242409928889181, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.02070560688285211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 42.583, - "cuda_time_us": 18.496, - "pct_cuda_time": 0.2546349101763515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.16, - "pct_cuda_time": 0.22247513778383649, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.032159772392514975, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 18.005, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 111.807, - "cuda_time_us": 132.38400000000001, - "pct_cuda_time": 1.8225339505182805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.12, - "cuda_time_us": 80.48, - "pct_cuda_time": 1.1079702406462353, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.48, - "pct_cuda_time": 1.1079702406462353, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.671, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12291200681522849, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12291200681522849, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 40.38, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.5916517030568167, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.5916517030568167, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 519.633, - "cuda_time_us": 212.48000000000002, - "pct_cuda_time": 2.925217653236979, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.986, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.04713829652053565, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.424, - "pct_cuda_time": 0.04713829652053565, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 371.717, - "cuda_time_us": 72.896, - "pct_cuda_time": 1.0035611165773852, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 29.835, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.28547304808698226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.28547304808698226, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 109.734, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242483444807236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.05242483444807236, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.878, - "cuda_time_us": 30.272, - "pct_cuda_time": 0.41675540662081056, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.904, - "pct_cuda_time": 0.32908698598916014, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.72, - "pct_cuda_time": 0.03744631032005169, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.01850288274637848, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 35.713, - "cuda_time_us": 18.08, - "pct_cuda_time": 0.24890782742152, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.968, - "pct_cuda_time": 0.2198318688200681, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.917, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 98.388, - "cuda_time_us": 132.96, - "pct_cuda_time": 1.8304637574095857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 34.382, - "cuda_time_us": 79.808, - "pct_cuda_time": 1.098718799273046, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.808, - "pct_cuda_time": 1.098718799273046, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 22.032, - "cuda_time_us": 9.152, - "pct_cuda_time": 0.12599582060629155, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.152, - "pct_cuda_time": 0.12599582060629155, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 29.988, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.6057491375302478, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.6057491375302478, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 561.599, - "cuda_time_us": 212.642, - "pct_cuda_time": 2.9274479114251584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.219, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.0462572068659462, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 406.506, - "cuda_time_us": 72.35300000000001, - "pct_cuda_time": 0.996085621539228, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 30.312, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.2823892342959192, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.2823892342959192, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 111.067, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05066265513889345, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 171.233, - "cuda_time_us": 29.793, - "pct_cuda_time": 0.41016100123724264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.744, - "pct_cuda_time": 0.3268842618526865, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.433, - "pct_cuda_time": 0.03349517390025211, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.018062337919083758, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 53.001, - "cuda_time_us": 18.368000000000002, - "pct_cuda_time": 0.2528727308671726, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.256, - "pct_cuda_time": 0.22379677226572067, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 23.287, - "cuda_time_us": 3.201, - "pct_cuda_time": 0.044068249755325535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.201, - "pct_cuda_time": 0.044068249755325535, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 101.384, - "cuda_time_us": 133.72799999999998, - "pct_cuda_time": 1.8410368332646583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 37.71, - "cuda_time_us": 80.416, - "pct_cuda_time": 1.1070891509916458, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.416, - "pct_cuda_time": 1.1070891509916458, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 21.797, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511473095170209, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.088, - "pct_cuda_time": 0.12511473095170209, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.145, - "cuda_time_us": 44.224, - "pct_cuda_time": 0.6088329513213109, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.224, - "pct_cuda_time": 0.6088329513213109, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 508.474, - "cuda_time_us": 213.56699999999998, - "pct_cuda_time": 2.9401824103391463, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.451, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.044935572384062025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.044935572384062025, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 368.469, - "cuda_time_us": 72.8, - "pct_cuda_time": 1.002239482095501, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 28.772, - "cuda_time_us": 21.088, - "pct_cuda_time": 0.29031904118722424, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.088, - "pct_cuda_time": 0.29031904118722424, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 106.636, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05022211031159874, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 164.669, - "cuda_time_us": 29.919999999999998, - "pct_cuda_time": 0.41190941352056853, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.648, - "pct_cuda_time": 0.3255626273708023, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.02070560688285211, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 33.578, - "cuda_time_us": 18.144, - "pct_cuda_time": 0.24978891707610945, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.0, - "pct_cuda_time": 0.22027241364736286, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.029516503428746628, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 16.081, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.043613937902177845, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 93.2, - "cuda_time_us": 134.33499999999998, - "pct_cuda_time": 1.8493934179574054, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 32.091, - "cuda_time_us": 80.991, - "pct_cuda_time": 1.115005190857098, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.991, - "pct_cuda_time": 1.115005190857098, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.217, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.12467418612440737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.12467418612440737, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 30.104, - "cuda_time_us": 44.288, - "pct_cuda_time": 0.6097140409759003, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.288, - "pct_cuda_time": 0.6097140409759003, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 498.637, - "cuda_time_us": 210.719, - "pct_cuda_time": 2.900973920709916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 13.998, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04581666203865147, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 359.737, - "cuda_time_us": 71.936, - "pct_cuda_time": 0.9903447717585435, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 27.29, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841514136050981, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.2841514136050981, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 104.745, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.05330592410266181, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.05330592410266181, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 160.366, - "cuda_time_us": 29.984999999999996, - "pct_cuda_time": 0.4128042702010109, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03171922756522025, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.68, - "pct_cuda_time": 0.32600317219809705, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.033921951701693875, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.537, - "pct_cuda_time": 0.021159918735999792, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 129], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 38.994, - "cuda_time_us": 17.439, - "pct_cuda_time": 0.24008316384977255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.327, - "pct_cuda_time": 0.21100720524832067, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.0290759586014519, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 15.598, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04405448272947257, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 94.822, - "cuda_time_us": 132.255, - "pct_cuda_time": 1.8207580041832483, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 33.853, - "cuda_time_us": 79.743, - "pct_cuda_time": 1.0978239425926035, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.743, - "pct_cuda_time": 1.0978239425926035, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 20.863, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.11982819302416539, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.11982819302416539, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 28.998, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6031058685664794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6031058685664794, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 14.619, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845993100241983, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.04845993100241983, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 95.484, - "cuda_time_us": 348.28700000000003, - "pct_cuda_time": 4.794876133249942, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04317339307488312, - "trace": "index_select(bfloat16[2, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010132531027778691, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 344.415, - "pct_cuda_time": 4.74157020914728, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 714.217, - "cuda_time_us": 124.543, - "pct_cuda_time": 1.7145867008052196, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03524358618357806, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.030397593083336075, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.0308381379106308, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.030397593083336075, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.175, - "pct_cuda_time": 0.029943281230188388, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.029957048256041348, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.177, - "pct_cuda_time": 0.029970815281894308, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.159, - "pct_cuda_time": 0.05725706052246138, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.768, - "pct_cuda_time": 0.06564117926691414, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.176, - "pct_cuda_time": 0.4705018755507671, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.648, - "pct_cuda_time": 0.38063073078264303, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 2.016, - "pct_cuda_time": 0.02775432411956772, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.896, - "pct_cuda_time": 0.06740335857609303, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 28.384, - "pct_cuda_time": 0.39076326181042176, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.752, - "pct_cuda_time": 0.03788685514734641, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - } -} \ No newline at end of file diff --git a/H100_llama8b_pp1_tp1/profiling_bs2_pl256.json b/H100_llama8b_pp1_tp1/profiling_bs2_pl256.json deleted file mode 100644 index b968481f1fd65101bc2a37e67edb512ff2addbc7..0000000000000000000000000000000000000000 --- a/H100_llama8b_pp1_tp1/profiling_bs2_pl256.json +++ /dev/null @@ -1,17561 +0,0 @@ -{ - "context": { - "python_version": "3.12.9 | packaged by Anaconda, Inc. | (main, Feb 6 2025, 18:56:27) [GCC 11.2.0]", - "torch_version": "2.5.1+cu124", - "engine_args": { - "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "served_model_name": null, - "tokenizer": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "task": "auto", - "skip_tokenizer_init": false, - "tokenizer_mode": "auto", - "trust_remote_code": false, - "allowed_local_media_path": null, - "download_dir": null, - "load_format": "dummy", - "config_format": "auto", - "dtype": "auto", - "kv_cache_dtype": "auto", - "seed": 0, - "max_model_len": null, - "distributed_executor_backend": null, - "pipeline_parallel_size": 1, - "tensor_parallel_size": 1, - "max_parallel_loading_workers": null, - "block_size": null, - "enable_prefix_caching": false, - "disable_sliding_window": false, - "use_v2_block_manager": true, - "swap_space": 4, - "cpu_offload_gb": 0, - "gpu_memory_utilization": 0.9, - "max_num_batched_tokens": 8000, - "max_num_partial_prefills": 1, - "max_long_partial_prefills": 1, - "long_prefill_token_threshold": 0, - "max_num_seqs": 256, - "max_logprobs": 20, - "disable_log_stats": false, - "revision": null, - "code_revision": null, - "rope_scaling": null, - "rope_theta": null, - "hf_overrides": null, - "tokenizer_revision": null, - "quantization": null, - "enforce_eager": true, - "max_seq_len_to_capture": 8192, - "disable_custom_all_reduce": false, - "tokenizer_pool_size": 0, - "tokenizer_pool_type": "ray", - "tokenizer_pool_extra_config": null, - "limit_mm_per_prompt": null, - "mm_processor_kwargs": null, - "disable_mm_preprocessor_cache": false, - "enable_lora": false, - "enable_lora_bias": false, - "max_loras": 1, - "max_lora_rank": 16, - "enable_prompt_adapter": false, - "max_prompt_adapters": 1, - "max_prompt_adapter_token": 0, - "fully_sharded_loras": false, - "lora_extra_vocab_size": 256, - "long_lora_scaling_factors": null, - "lora_dtype": "auto", - "max_cpu_loras": null, - "device": "auto", - "num_scheduler_steps": 1, - "multi_step_stream_outputs": true, - "ray_workers_use_nsight": false, - "num_gpu_blocks_override": null, - "num_lookahead_slots": 0, - "model_loader_extra_config": null, - "ignore_patterns": [], - "preemption_mode": null, - "scheduler_delay_factor": 0.0, - "enable_chunked_prefill": null, - "guided_decoding_backend": "xgrammar", - "logits_processor_pattern": null, - "speculative_model": null, - "speculative_model_quantization": null, - "speculative_draft_tensor_parallel_size": null, - "num_speculative_tokens": null, - "speculative_disable_mqa_scorer": false, - "speculative_max_model_len": null, - "speculative_disable_by_batch_size": null, - "ngram_prompt_lookup_max": null, - "ngram_prompt_lookup_min": null, - "spec_decoding_acceptance_method": "rejection_sampler", - "typical_acceptance_sampler_posterior_threshold": null, - "typical_acceptance_sampler_posterior_alpha": null, - "qlora_adapter_name_or_path": null, - "disable_logprobs_during_spec_decoding": null, - "otlp_traces_endpoint": null, - "collect_detailed_traces": null, - "disable_async_output_proc": false, - "scheduling_policy": "fcfs", - "scheduler_cls": "vllm.core.scheduler.Scheduler", - "override_neuron_config": null, - "override_pooler_config": null, - "compilation_config": null, - "worker_cls": "auto", - "kv_transfer_config": null, - "generation_config": null, - "override_generation_config": null, - "enable_sleep_mode": false, - "model_impl": "auto", - "calculate_kv_scales": false, - "additional_config": null - }, - "prompt_len": 0, - "batch_size": 2, - "num_steps": 2, - "complete_num_requests_per_step": null, - "save_chrome_traces_folder": null - }, - "prefill": { - "metadata": { - "num_running_seqs": null - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 12663.174, - "pct_cuda_time": 96.3953153875426, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 16.48, - "pct_cuda_time": 0.12544996993539706, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cuda_time_us": 16.48, - "pct_cuda_time": 0.12544996993539706, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 12639.717999999999, - "pct_cuda_time": 96.21676232353742, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 437.1769999999999, - "pct_cuda_time": 3.3279030040319824, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 7.808, - "pct_cuda_time": 0.05943649061016871, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 429.36899999999997, - "pct_cuda_time": 3.2684665134218136, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 3257.473, - "pct_cuda_time": 24.796716621078136, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 1438.719, - "pct_cuda_time": 10.951896559192024, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1438.719, - "pct_cuda_time": 10.951896559192024, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 220.70300000000003, - "pct_cuda_time": 1.6800476161803364, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 220.70300000000003, - "pct_cuda_time": 1.6800476161803364, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 534.659, - "pct_cuda_time": 4.069960890515137, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 120.06400000000004, - "pct_cuda_time": 0.91395783921866, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 369.02600000000007, - "pct_cuda_time": 2.809120182365282, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 45.568999999999996, - "pct_cuda_time": 0.3468828689311959, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 1063.3919999999998, - "pct_cuda_time": 8.094811555190638, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1063.3919999999998, - "pct_cuda_time": 8.094811555190638, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 8945.068, - "pct_cuda_time": 68.09214269842731, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 5434.482000000001, - "pct_cuda_time": 41.368665261799556, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 24.76600000000001, - "pct_cuda_time": 0.18852511865412894, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 5409.716000000001, - "pct_cuda_time": 41.18014014314543, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 731.04, - "pct_cuda_time": 5.56486322946436, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 731.04, - "pct_cuda_time": 5.56486322946436, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 2779.546, - "pct_cuda_time": 21.158614207163417, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2779.546, - "pct_cuda_time": 21.158614207163417, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 354.71999999999997, - "pct_cuda_time": 2.7002192557939346, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.072, - "pct_cuda_time": 0.02338484876465654, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 350.912, - "pct_cuda_time": 2.6712317870127458, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 118.817, - "pct_cuda_time": 0.9044653566634753, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 11.137000000000002, - "pct_cuda_time": 0.0847776890273372, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.096, - "pct_cuda_time": 0.031179798352875387, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.64, - "pct_cuda_time": 0.03532086532161665, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 34.016, - "pct_cuda_time": 0.2589384816336448, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.776, - "pct_cuda_time": 0.21143800758043624, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 2.048, - "pct_cuda_time": 0.015589899176437693, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.832, - "pct_cuda_time": 0.03678241836940768, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 28.0, - "pct_cuda_time": 0.2131431528028591, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.272, - "pct_cuda_time": 0.017295044398860565, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 33540.417, - "cuda_time_us": 12663.174, - "pct_cuda_time": 96.3953153875426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 107.841, - "cuda_time_us": 16.48, - "pct_cuda_time": 0.12544996993539706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 16.48, - "pct_cuda_time": 0.12544996993539706, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[512]) <- embedding(bfloat16[128256, 4096], int64[512], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1516.87, - "cuda_time_us": 389.407, - "pct_cuda_time": 2.964265560839391, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 106.378, - "cuda_time_us": 7.808, - "pct_cuda_time": 0.05943649061016871, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.808, - "pct_cuda_time": 0.05943649061016871, - "trace": "_C::rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 1097.382, - "cuda_time_us": 102.78399999999999, - "pct_cuda_time": 0.7824180649174667, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 128.047, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.34200341318310185, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.34200341318310185, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 369.406, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 399.887, - "cuda_time_us": 16.96, - "pct_cuda_time": 0.12910385255487464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.936, - "pct_cuda_time": 0.029961837479716192, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.08891114374062122, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.010230871334537237, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 105.477, - "cuda_time_us": 34.24, - "pct_cuda_time": 0.2606436268560677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 34.24, - "pct_cuda_time": 0.2606436268560677, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 42.802, - "cuda_time_us": 6.528, - "pct_cuda_time": 0.04969280362489514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.528, - "pct_cuda_time": 0.04969280362489514, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 218.532, - "cuda_time_us": 272.287, - "pct_cuda_time": 2.0727182016868606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 75.066, - "cuda_time_us": 170.591, - "pct_cuda_time": 1.2985822707068762, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.005595007761075051, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.856, - "pct_cuda_time": 1.2929872629458012, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 52.75, - "cuda_time_us": 22.592, - "pct_cuda_time": 0.17197607529007827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.592, - "pct_cuda_time": 0.17197607529007827, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 60.07, - "cuda_time_us": 79.104, - "pct_cuda_time": 0.6021598556899058, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.104, - "pct_cuda_time": 0.6021598556899058, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 990.64, - "cuda_time_us": 399.135, - "pct_cuda_time": 3.03831758192747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.29, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 711.857, - "cuda_time_us": 102.207, - "pct_cuda_time": 0.7780257935186363, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.908, - "cuda_time_us": 45.76, - "pct_cuda_time": 0.34833680972352965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.76, - "pct_cuda_time": 0.34833680972352965, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 203.454, - "cuda_time_us": 6.944, - "pct_cuda_time": 0.05285950189510906, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.944, - "pct_cuda_time": 0.05285950189510906, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 313.993, - "cuda_time_us": 16.64, - "pct_cuda_time": 0.12666793080855626, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.02801310008266148, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.08720599851819834, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01144883220769643, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 74.493, - "cuda_time_us": 32.863, - "pct_cuda_time": 0.25016155109144134, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.863, - "pct_cuda_time": 0.25016155109144134, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.558, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 184.928, - "cuda_time_us": 283.64799999999997, - "pct_cuda_time": 2.1592010359366203, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.966, - "cuda_time_us": 170.112, - "pct_cuda_time": 1.2949360003428558, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.344, - "pct_cuda_time": 1.2890897881516916, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 42.143, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.17270685181397383, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.17270685181397383, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.947, - "cuda_time_us": 90.848, - "pct_cuda_time": 0.6915581837797908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 90.848, - "pct_cuda_time": 0.6915581837797908, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1012.761, - "cuda_time_us": 401.79299999999995, - "pct_cuda_time": 3.058550956932827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.717, - "cuda_time_us": 6.72, - "pct_cuda_time": 0.051154356672686176, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.72, - "pct_cuda_time": 0.051154356672686176, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 714.587, - "cuda_time_us": 103.10499999999999, - "pct_cuda_time": 0.7848615989192422, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.653, - "cuda_time_us": 44.448, - "pct_cuda_time": 0.3383495305636243, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.448, - "pct_cuda_time": 0.3383495305636243, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 195.496, - "cuda_time_us": 7.296, - "pct_cuda_time": 0.05553901581605928, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.296, - "pct_cuda_time": 0.05553901581605928, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 323.942, - "cuda_time_us": 16.833000000000002, - "pct_cuda_time": 0.12813709611180454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.713, - "pct_cuda_time": 0.028264304512750565, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.08842395939135754, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01144883220769643, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 78.648, - "cuda_time_us": 34.528, - "pct_cuda_time": 0.2628359564277542, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 34.528, - "pct_cuda_time": 0.2628359564277542, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 34.798, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 195.146, - "cuda_time_us": 284.99199999999996, - "pct_cuda_time": 2.169431907271157, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 71.32, - "cuda_time_us": 168.608, - "pct_cuda_time": 1.2834871681351594, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 167.872, - "pct_cuda_time": 1.2778845481186272, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 41.815, - "cuda_time_us": 23.104, - "pct_cuda_time": 0.1758735500841877, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 23.104, - "pct_cuda_time": 0.1758735500841877, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 57.384, - "cuda_time_us": 93.28, - "pct_cuda_time": 0.7100711890518105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 93.28, - "pct_cuda_time": 0.7100711890518105, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 961.489, - "cuda_time_us": 386.46299999999997, - "pct_cuda_time": 2.9418550807732613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.218, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 692.574, - "cuda_time_us": 100.22399999999999, - "pct_cuda_time": 0.7629306909469195, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.643, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.3351828322934104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.3351828322934104, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 194.414, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 307.17, - "cuda_time_us": 16.736, - "pct_cuda_time": 0.1273987073324518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 4.0, - "pct_cuda_time": 0.03044902182897987, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.424, - "pct_cuda_time": 0.0869624063435665, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.009987279159905398, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.733, - "cuda_time_us": 32.672, - "pct_cuda_time": 0.24870761029910757, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.672, - "pct_cuda_time": 0.24870761029910757, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.209, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.237, - "cuda_time_us": 272.51099999999997, - "pct_cuda_time": 2.074423346909283, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 64.29, - "cuda_time_us": 169.727, - "pct_cuda_time": 1.2920052819918166, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.991, - "pct_cuda_time": 1.2864026619752844, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.899, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.17221966746471012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.17221966746471012, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.881, - "cuda_time_us": 80.16, - "pct_cuda_time": 0.6101983974527565, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.16, - "pct_cuda_time": 0.6101983974527565, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 928.5, - "cuda_time_us": 386.721, - "pct_cuda_time": 2.943819042681231, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.689, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 657.775, - "cuda_time_us": 101.40899999999999, - "pct_cuda_time": 0.7719512136637549, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.884, - "cuda_time_us": 45.792, - "pct_cuda_time": 0.3485804018981616, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.792, - "pct_cuda_time": 0.3485804018981616, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 178.683, - "cuda_time_us": 6.592, - "pct_cuda_time": 0.05017998797415882, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.592, - "pct_cuda_time": 0.05017998797415882, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 297.52, - "cuda_time_us": 16.480999999999998, - "pct_cuda_time": 0.1254575821908543, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.02825669225729332, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.08720599851819834, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.313, - "pct_cuda_time": 0.00999489141536264, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.622, - "cuda_time_us": 32.544, - "pct_cuda_time": 0.2477332416005802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.544, - "pct_cuda_time": 0.2477332416005802, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 34.436, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 180.991, - "cuda_time_us": 271.552, - "pct_cuda_time": 2.0671231939257853, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.206, - "cuda_time_us": 169.504, - "pct_cuda_time": 1.290307749024851, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.736, - "pct_cuda_time": 1.2844615368336867, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.953, - "cuda_time_us": 22.815, - "pct_cuda_time": 0.17367360825704395, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.815, - "pct_cuda_time": 0.17367360825704395, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 58.407, - "cuda_time_us": 79.233, - "pct_cuda_time": 0.6031418366438905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.233, - "pct_cuda_time": 0.6031418366438905, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1051.938, - "cuda_time_us": 394.017, - "pct_cuda_time": 2.99935805849729, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.769, - "cuda_time_us": 7.105, - "pct_cuda_time": 0.0540850750237255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.105, - "pct_cuda_time": 0.0540850750237255, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 749.369, - "cuda_time_us": 102.401, - "pct_cuda_time": 0.7795025710773419, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.688, - "cuda_time_us": 45.728, - "pct_cuda_time": 0.3480932175488979, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.728, - "pct_cuda_time": 0.3480932175488979, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 250.479, - "cuda_time_us": 7.264, - "pct_cuda_time": 0.05529542364142745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.264, - "pct_cuda_time": 0.05529542364142745, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 306.778, - "cuda_time_us": 16.769, - "pct_cuda_time": 0.12764991176254084, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.028987468781188835, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.08720599851819834, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.011456444463153675, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.455, - "cuda_time_us": 32.64, - "pct_cuda_time": 0.24846401812447572, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.64, - "pct_cuda_time": 0.24846401812447572, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.131, - "cuda_time_us": 6.432, - "pct_cuda_time": 0.04896202710099963, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.432, - "pct_cuda_time": 0.04896202710099963, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 185.123, - "cuda_time_us": 278.079, - "pct_cuda_time": 2.1168083852952235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 67.225, - "cuda_time_us": 169.375, - "pct_cuda_time": 1.2893257680708663, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.639, - "pct_cuda_time": 1.283723148054334, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 41.096, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.17221966746471012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.17221966746471012, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.466, - "cuda_time_us": 86.08, - "pct_cuda_time": 0.6552629497596467, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 86.08, - "pct_cuda_time": 0.6552629497596467, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 916.988, - "cuda_time_us": 397.119, - "pct_cuda_time": 3.0229712749256645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.485, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 658.464, - "cuda_time_us": 101.119, - "pct_cuda_time": 0.7697436595811539, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.103, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.34175982100847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.34175982100847, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.473, - "cuda_time_us": 6.719, - "pct_cuda_time": 0.05114674441722894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.719, - "pct_cuda_time": 0.05114674441722894, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 296.663, - "cuda_time_us": 16.672, - "pct_cuda_time": 0.1269115229831881, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.02776950790802964, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.52, - "pct_cuda_time": 0.08769318286746201, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01144883220769643, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 67.679, - "cuda_time_us": 32.832, - "pct_cuda_time": 0.2499255711722668, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.832, - "pct_cuda_time": 0.2499255711722668, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.345, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 174.699, - "cuda_time_us": 282.464, - "pct_cuda_time": 2.1501881254752426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.823, - "cuda_time_us": 170.72, - "pct_cuda_time": 1.2995642516608608, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.696, - "pct_cuda_time": 0.012910385255487464, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.024, - "pct_cuda_time": 1.2866538664053735, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.318, - "cuda_time_us": 22.528, - "pct_cuda_time": 0.17148889094081463, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.528, - "pct_cuda_time": 0.17148889094081463, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.263, - "cuda_time_us": 89.216, - "pct_cuda_time": 0.679134982873567, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 89.216, - "pct_cuda_time": 0.679134982873567, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 958.889, - "cuda_time_us": 392.543, - "pct_cuda_time": 2.9881375939533115, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.048, - "cuda_time_us": 6.848, - "pct_cuda_time": 0.05212872537121353, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.848, - "pct_cuda_time": 0.05212872537121353, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 682.236, - "cuda_time_us": 100.032, - "pct_cuda_time": 0.7614691378991285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.481, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.3407854523099427, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.3407854523099427, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.579, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 310.393, - "cuda_time_us": 16.448, - "pct_cuda_time": 0.1252063777607652, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.028500284431925156, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.392, - "pct_cuda_time": 0.08671881416893466, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.009987279159905398, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 75.51, - "cuda_time_us": 32.032, - "pct_cuda_time": 0.24383576680647076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.032, - "pct_cuda_time": 0.24383576680647076, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.972, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 184.599, - "cuda_time_us": 279.039, - "pct_cuda_time": 2.1241161505341783, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 64.965, - "cuda_time_us": 169.72799999999998, - "pct_cuda_time": 1.2920128942472737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.992, - "pct_cuda_time": 1.2864102742307415, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.972, - "cuda_time_us": 23.328, - "pct_cuda_time": 0.1775786953066106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 23.328, - "pct_cuda_time": 0.1775786953066106, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.311, - "cuda_time_us": 85.983, - "pct_cuda_time": 0.6545245609802941, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 85.983, - "pct_cuda_time": 0.6545245609802941, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 972.444, - "cuda_time_us": 400.03000000000003, - "pct_cuda_time": 3.0451305505617046, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.373, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 665.338, - "cuda_time_us": 102.207, - "pct_cuda_time": 0.7780257935186363, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 63.116, - "cuda_time_us": 45.056, - "pct_cuda_time": 0.34297778188162925, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.056, - "pct_cuda_time": 0.34297778188162925, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.143, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 290.068, - "cuda_time_us": 16.671, - "pct_cuda_time": 0.12690391072773083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.02825669225729332, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.08842395939135754, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.010223259079079991, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 69.392, - "cuda_time_us": 33.472, - "pct_cuda_time": 0.2547974146649036, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.472, - "pct_cuda_time": 0.2547974146649036, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 40.607, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 207.432, - "cuda_time_us": 284.19100000000003, - "pct_cuda_time": 2.163334490649905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.465, - "cuda_time_us": 171.232, - "pct_cuda_time": 1.3034617264549702, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 170.496, - "pct_cuda_time": 1.2978591064384382, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 40.158, - "cuda_time_us": 22.591, - "pct_cuda_time": 0.17196846303462107, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.591, - "pct_cuda_time": 0.17196846303462107, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 84.936, - "cuda_time_us": 90.368, - "pct_cuda_time": 0.6879043011603131, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 90.368, - "pct_cuda_time": 0.6879043011603131, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1051.659, - "cuda_time_us": 398.46299999999997, - "pct_cuda_time": 3.0332021462602015, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.315, - "cuda_time_us": 6.944, - "pct_cuda_time": 0.05285950189510906, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.944, - "pct_cuda_time": 0.05285950189510906, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 728.16, - "cuda_time_us": 101.505, - "pct_cuda_time": 0.7726819901876504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.535, - "cuda_time_us": 44.864, - "pct_cuda_time": 0.3415162288338382, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.864, - "pct_cuda_time": 0.3415162288338382, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 181.268, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 339.53, - "cuda_time_us": 16.801, - "pct_cuda_time": 0.12789350393717266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.02825669225729332, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.0881803672167257, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.011456444463153675, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 90.626, - "cuda_time_us": 33.216, - "pct_cuda_time": 0.2528486772678488, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.216, - "pct_cuda_time": 0.2528486772678488, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 39.643, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 223.671, - "cuda_time_us": 283.22999999999996, - "pct_cuda_time": 2.1560191131554918, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 79.557, - "cuda_time_us": 169.951, - "pct_cuda_time": 1.2937104272142392, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.215, - "pct_cuda_time": 1.288107807197707, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 47.4, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.17270685181397383, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.17270685181397383, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 67.594, - "cuda_time_us": 90.591, - "pct_cuda_time": 0.6896018341272788, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 90.591, - "pct_cuda_time": 0.6896018341272788, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1160.956, - "cuda_time_us": 400.413, - "pct_cuda_time": 3.048046044401829, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 34.504, - "cuda_time_us": 6.911, - "pct_cuda_time": 0.05260829746501996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.911, - "pct_cuda_time": 0.05260829746501996, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 841.621, - "cuda_time_us": 101.599, - "pct_cuda_time": 0.7733975422006315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 66.929, - "cuda_time_us": 44.543, - "pct_cuda_time": 0.3390726948320626, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.543, - "pct_cuda_time": 0.3390726948320626, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 239.607, - "cuda_time_us": 6.944, - "pct_cuda_time": 0.05285950189510906, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.944, - "pct_cuda_time": 0.05285950189510906, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 384.02, - "cuda_time_us": 16.8, - "pct_cuda_time": 0.12788589168171546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.02801310008266148, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.616, - "pct_cuda_time": 0.08842395939135754, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01144883220769643, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 81.743, - "cuda_time_us": 33.312, - "pct_cuda_time": 0.25357945379174435, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.312, - "pct_cuda_time": 0.25357945379174435, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 37.473, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 213.29, - "cuda_time_us": 285.087, - "pct_cuda_time": 2.170155071539596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 76.968, - "cuda_time_us": 169.40699999999998, - "pct_cuda_time": 1.289569360245498, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.671, - "pct_cuda_time": 1.2839667402289658, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 44.197, - "cuda_time_us": 22.976, - "pct_cuda_time": 0.17489918138566035, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.976, - "pct_cuda_time": 0.17489918138566035, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 68.732, - "cuda_time_us": 92.704, - "pct_cuda_time": 0.7056865299084374, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 92.704, - "pct_cuda_time": 0.7056865299084374, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1170.874, - "cuda_time_us": 386.976, - "pct_cuda_time": 2.9457601678228285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.18, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 855.473, - "cuda_time_us": 100.8, - "pct_cuda_time": 0.7673153500902927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 66.691, - "cuda_time_us": 44.256, - "pct_cuda_time": 0.33688797751583327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.256, - "pct_cuda_time": 0.33688797751583327, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 224.256, - "cuda_time_us": 7.232, - "pct_cuda_time": 0.055051831466795605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.232, - "pct_cuda_time": 0.055051831466795605, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 364.226, - "cuda_time_us": 16.544, - "pct_cuda_time": 0.12593715428466076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.028743876606556997, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.488, - "pct_cuda_time": 0.08744959069283018, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.009743686985273559, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 82.366, - "cuda_time_us": 32.768, - "pct_cuda_time": 0.2494383868230031, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.768, - "pct_cuda_time": 0.2494383868230031, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 35.619, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 213.099, - "cuda_time_us": 272.608, - "pct_cuda_time": 2.0751617356886363, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 76.003, - "cuda_time_us": 169.216, - "pct_cuda_time": 1.2881154194531645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.448, - "pct_cuda_time": 1.2822692072620003, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 45.355, - "cuda_time_us": 23.68, - "pct_cuda_time": 0.18025820922756083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 23.68, - "pct_cuda_time": 0.18025820922756083, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 66.722, - "cuda_time_us": 79.712, - "pct_cuda_time": 0.6067881070079109, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.712, - "pct_cuda_time": 0.6067881070079109, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1287.147, - "cuda_time_us": 402.782, - "pct_cuda_time": 3.0660794775800424, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.085, - "cuda_time_us": 7.072, - "pct_cuda_time": 0.05383387059363641, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.072, - "pct_cuda_time": 0.05383387059363641, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 939.373, - "cuda_time_us": 102.464, - "pct_cuda_time": 0.7799821431711483, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 64.585, - "cuda_time_us": 45.76, - "pct_cuda_time": 0.34833680972352965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.76, - "pct_cuda_time": 0.34833680972352965, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 234.015, - "cuda_time_us": 6.528, - "pct_cuda_time": 0.04969280362489514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.528, - "pct_cuda_time": 0.04969280362489514, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 476.816, - "cuda_time_us": 16.544, - "pct_cuda_time": 0.12593715428466076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.02801310008266148, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.488, - "pct_cuda_time": 0.08744959069283018, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.376, - "pct_cuda_time": 0.010474463509169075, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 94.184, - "cuda_time_us": 33.632, - "pct_cuda_time": 0.25601537553806275, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.632, - "pct_cuda_time": 0.25601537553806275, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 41.709, - "cuda_time_us": 6.496, - "pct_cuda_time": 0.049449211450263306, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.496, - "pct_cuda_time": 0.049449211450263306, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 239.226, - "cuda_time_us": 286.75, - "pct_cuda_time": 2.1828142523649943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 79.948, - "cuda_time_us": 170.238, - "pct_cuda_time": 1.2958951445304687, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.005595007761075051, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.503, - "pct_cuda_time": 1.2903001367693936, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 59.62, - "cuda_time_us": 23.04, - "pct_cuda_time": 0.17538636573492403, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 23.04, - "pct_cuda_time": 0.17538636573492403, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 70.181, - "cuda_time_us": 93.472, - "pct_cuda_time": 0.7115327420996015, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 93.472, - "pct_cuda_time": 0.7115327420996015, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1130.803, - "cuda_time_us": 395.83899999999994, - "pct_cuda_time": 3.0132275879403903, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.902, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 798.424, - "cuda_time_us": 100.768, - "pct_cuda_time": 0.7670717579156608, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 66.453, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.3376187540397288, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.3376187540397288, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 219.939, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 363.193, - "cuda_time_us": 16.896, - "pct_cuda_time": 0.128616668205611, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.028987468781188835, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.0881803672167257, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01144883220769643, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 80.031, - "cuda_time_us": 32.512, - "pct_cuda_time": 0.2474896494259484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.512, - "pct_cuda_time": 0.2474896494259484, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 50.258, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 215.11, - "cuda_time_us": 281.53499999999997, - "pct_cuda_time": 2.1431163401554616, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 75.459, - "cuda_time_us": 170.111, - "pct_cuda_time": 1.2949283880873985, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.375, - "pct_cuda_time": 1.2893257680708663, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 47.53, - "cuda_time_us": 22.56, - "pct_cuda_time": 0.17173248311544645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.56, - "pct_cuda_time": 0.17173248311544645, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 66.194, - "cuda_time_us": 88.864, - "pct_cuda_time": 0.6764554689526168, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 88.864, - "pct_cuda_time": 0.6764554689526168, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1137.554, - "cuda_time_us": 397.181, - "pct_cuda_time": 3.023443234764013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.506, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 818.622, - "cuda_time_us": 101.79, - "pct_cuda_time": 0.7748514829929652, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 63.843, - "cuda_time_us": 44.736, - "pct_cuda_time": 0.34054186013531085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.736, - "pct_cuda_time": 0.34054186013531085, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 222.486, - "cuda_time_us": 7.2, - "pct_cuda_time": 0.05480823929216377, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.2, - "pct_cuda_time": 0.05480823929216377, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 380.233, - "cuda_time_us": 16.607, - "pct_cuda_time": 0.12641672637846718, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.02776950790802964, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.487, - "pct_cuda_time": 0.08744197843737293, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.011205240033064591, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 82.23, - "cuda_time_us": 33.247, - "pct_cuda_time": 0.2530846571870234, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.247, - "pct_cuda_time": 0.2530846571870234, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 36.357, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 215.068, - "cuda_time_us": 281.823, - "pct_cuda_time": 2.1453086697271484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 75.969, - "cuda_time_us": 169.63199999999998, - "pct_cuda_time": 1.2912821177233782, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.896, - "pct_cuda_time": 1.285679497706846, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 45.996, - "cuda_time_us": 23.456, - "pct_cuda_time": 0.17855306400513796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 23.456, - "pct_cuda_time": 0.17855306400513796, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 66.327, - "cuda_time_us": 88.735, - "pct_cuda_time": 0.6754734879986322, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 88.735, - "pct_cuda_time": 0.6754734879986322, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1139.27, - "cuda_time_us": 403.1329999999999, - "pct_cuda_time": 3.0687513792455348, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.316, - "cuda_time_us": 7.136, - "pct_cuda_time": 0.05432105494290009, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.136, - "pct_cuda_time": 0.05432105494290009, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 822.263, - "cuda_time_us": 102.20599999999999, - "pct_cuda_time": 0.778018181263179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 65.126, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.3407854523099427, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.3407854523099427, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 227.624, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 377.115, - "cuda_time_us": 16.894, - "pct_cuda_time": 0.12860144369469645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.968, - "pct_cuda_time": 0.030205429654348033, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.615, - "pct_cuda_time": 0.0884163471359003, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.311, - "pct_cuda_time": 0.009979666904448152, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 80.158, - "cuda_time_us": 33.856, - "pct_cuda_time": 0.2577205207604856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.856, - "pct_cuda_time": 0.2577205207604856, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 36.205, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 212.992, - "cuda_time_us": 286.97499999999997, - "pct_cuda_time": 2.184527009842874, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 75.213, - "cuda_time_us": 170.688, - "pct_cuda_time": 1.2993206594862288, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.952, - "pct_cuda_time": 1.2937180394696968, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 46.982, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.17270685181397383, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.17270685181397383, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 66.59, - "cuda_time_us": 93.599, - "pct_cuda_time": 0.7124994985426718, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 93.599, - "pct_cuda_time": 0.7124994985426718, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1030.29, - "cuda_time_us": 398.11, - "pct_cuda_time": 3.030515020083794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.378, - "cuda_time_us": 7.039, - "pct_cuda_time": 0.053582666163547316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.039, - "pct_cuda_time": 0.053582666163547316, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 767.535, - "cuda_time_us": 100.12800000000001, - "pct_cuda_time": 0.7621999144230243, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 66.033, - "cuda_time_us": 44.448, - "pct_cuda_time": 0.3383495305636243, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.448, - "pct_cuda_time": 0.3383495305636243, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 220.784, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 346.204, - "cuda_time_us": 16.864, - "pct_cuda_time": 0.12837307603097914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.028987468781188835, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.712, - "pct_cuda_time": 0.08915473591525305, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.010230871334537237, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 70.917, - "cuda_time_us": 32.064, - "pct_cuda_time": 0.2440793589811026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.064, - "pct_cuda_time": 0.2440793589811026, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.473, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.872, - "cuda_time_us": 284.255, - "pct_cuda_time": 2.163821674999168, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.155, - "cuda_time_us": 170.39999999999998, - "pct_cuda_time": 1.2971283299145422, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.664, - "pct_cuda_time": 1.29152570989801, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.869, - "cuda_time_us": 22.56, - "pct_cuda_time": 0.17173248311544645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.56, - "pct_cuda_time": 0.17173248311544645, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.936, - "cuda_time_us": 91.295, - "pct_cuda_time": 0.6949608619691793, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 91.295, - "pct_cuda_time": 0.6949608619691793, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 941.438, - "cuda_time_us": 387.039, - "pct_cuda_time": 2.946239739916635, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.856, - "cuda_time_us": 6.879, - "pct_cuda_time": 0.05236470529038813, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.879, - "pct_cuda_time": 0.05236470529038813, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 671.26, - "cuda_time_us": 101.888, - "pct_cuda_time": 0.7755974840277753, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.621, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.3376187540397288, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.3376187540397288, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 179.412, - "cuda_time_us": 7.68, - "pct_cuda_time": 0.05846212191164135, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.68, - "pct_cuda_time": 0.05846212191164135, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 308.162, - "cuda_time_us": 16.864, - "pct_cuda_time": 0.12837307603097914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.872, - "pct_cuda_time": 0.029474653130452513, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.487, - "pct_cuda_time": 0.08744197843737293, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.011456444463153675, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.234, - "cuda_time_us": 32.992, - "pct_cuda_time": 0.25114353204542594, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.992, - "pct_cuda_time": 0.25114353204542594, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.011, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.061, - "cuda_time_us": 271.52, - "pct_cuda_time": 2.0668796017511535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.098, - "cuda_time_us": 168.832, - "pct_cuda_time": 1.2851923133575824, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.096, - "pct_cuda_time": 1.27958969334105, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 43.629, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.17221966746471012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.17221966746471012, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.476, - "cuda_time_us": 80.064, - "pct_cuda_time": 0.609467620928861, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.064, - "pct_cuda_time": 0.609467620928861, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 959.526, - "cuda_time_us": 394.046, - "pct_cuda_time": 2.99957881390555, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.497, - "cuda_time_us": 7.136, - "pct_cuda_time": 0.05432105494290009, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.136, - "pct_cuda_time": 0.05432105494290009, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 684.327, - "cuda_time_us": 103.295, - "pct_cuda_time": 0.7863079274561189, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.117, - "cuda_time_us": 46.239, - "pct_cuda_time": 0.35198308008755, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 46.239, - "pct_cuda_time": 0.35198308008755, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 198.982, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 302.254, - "cuda_time_us": 16.736, - "pct_cuda_time": 0.1273987073324518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.02776950790802964, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.488, - "pct_cuda_time": 0.08744959069283018, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.6, - "pct_cuda_time": 0.012179608731591948, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.654, - "cuda_time_us": 33.568, - "pct_cuda_time": 0.25552819118879905, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.568, - "pct_cuda_time": 0.25552819118879905, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 39.13, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.367, - "cuda_time_us": 276.799, - "pct_cuda_time": 2.1070646983099497, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 66.618, - "cuda_time_us": 169.535, - "pct_cuda_time": 1.2905437289440256, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.799, - "pct_cuda_time": 1.2849411089274934, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.801, - "cuda_time_us": 22.72, - "pct_cuda_time": 0.17295044398860565, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.72, - "pct_cuda_time": 0.17295044398860565, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.354, - "cuda_time_us": 84.544, - "pct_cuda_time": 0.6435705253773185, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 84.544, - "pct_cuda_time": 0.6435705253773185, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1090.085, - "cuda_time_us": 393.31199999999995, - "pct_cuda_time": 2.993991418399932, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.135, - "cuda_time_us": 7.136, - "pct_cuda_time": 0.05432105494290009, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.136, - "pct_cuda_time": 0.05432105494290009, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 809.353, - "cuda_time_us": 101.12, - "pct_cuda_time": 0.7697512718366112, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.901, - "cuda_time_us": 45.088, - "pct_cuda_time": 0.3432213740562611, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.088, - "pct_cuda_time": 0.3432213740562611, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 187.659, - "cuda_time_us": 6.528, - "pct_cuda_time": 0.04969280362489514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.528, - "pct_cuda_time": 0.04969280362489514, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 427.129, - "cuda_time_us": 16.544, - "pct_cuda_time": 0.12593715428466076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.02801310008266148, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.552, - "pct_cuda_time": 0.08793677504209385, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.009987279159905398, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 81.68, - "cuda_time_us": 32.96, - "pct_cuda_time": 0.2508999398707941, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.96, - "pct_cuda_time": 0.2508999398707941, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 35.243, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 186.637, - "cuda_time_us": 278.304, - "pct_cuda_time": 2.118521142773103, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 66.701, - "cuda_time_us": 169.56699999999998, - "pct_cuda_time": 1.2907873211186571, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.831, - "pct_cuda_time": 1.285184701102125, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 42.113, - "cuda_time_us": 22.945, - "pct_cuda_time": 0.17466320146648578, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.945, - "pct_cuda_time": 0.17466320146648578, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.251, - "cuda_time_us": 85.792, - "pct_cuda_time": 0.6530706201879602, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 85.792, - "pct_cuda_time": 0.6530706201879602, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 961.379, - "cuda_time_us": 395.22900000000004, - "pct_cuda_time": 3.0085841121114716, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 36.335, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 682.299, - "cuda_time_us": 100.99, - "pct_cuda_time": 0.7687616786271693, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.827, - "cuda_time_us": 45.056, - "pct_cuda_time": 0.34297778188162925, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.056, - "pct_cuda_time": 0.34297778188162925, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 196.33, - "cuda_time_us": 7.232, - "pct_cuda_time": 0.055051831466795605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.232, - "pct_cuda_time": 0.055051831466795605, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 294.014, - "cuda_time_us": 16.83, - "pct_cuda_time": 0.12811425934543277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.839, - "pct_cuda_time": 0.02922344870036343, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.584, - "pct_cuda_time": 0.0881803672167257, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.407, - "pct_cuda_time": 0.01071044342834367, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 75.281, - "cuda_time_us": 31.872, - "pct_cuda_time": 0.2426178059333116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 31.872, - "pct_cuda_time": 0.2426178059333116, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.252, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.159, - "cuda_time_us": 280.57500000000005, - "pct_cuda_time": 2.1358085749165068, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.034, - "cuda_time_us": 169.919, - "pct_cuda_time": 1.2934668350396077, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.151, - "pct_cuda_time": 1.2876206228484435, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.763, - "cuda_time_us": 22.657, - "pct_cuda_time": 0.17247087189479923, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.657, - "pct_cuda_time": 0.17247087189479923, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.793, - "cuda_time_us": 87.999, - "pct_cuda_time": 0.6698708679820998, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 87.999, - "pct_cuda_time": 0.6698708679820998, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 962.411, - "cuda_time_us": 398.559, - "pct_cuda_time": 3.033932922784097, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.994, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 670.756, - "cuda_time_us": 101.313, - "pct_cuda_time": 0.7712204371398593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 59.519, - "cuda_time_us": 44.32, - "pct_cuda_time": 0.3373751618650969, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.32, - "pct_cuda_time": 0.3373751618650969, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.203, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.624, - "pct_cuda_time": 0.05042358014879066, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 299.344, - "cuda_time_us": 16.961000000000002, - "pct_cuda_time": 0.1291114648103319, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.775, - "pct_cuda_time": 0.02873626435109975, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.489, - "pct_cuda_time": 0.08745720294828743, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.697, - "pct_cuda_time": 0.012917997510944709, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.549, - "cuda_time_us": 33.408, - "pct_cuda_time": 0.2543102303156399, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.408, - "pct_cuda_time": 0.2543102303156399, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 34.056, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 194.83, - "cuda_time_us": 283.61400000000003, - "pct_cuda_time": 2.1589422192510743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.226, - "cuda_time_us": 170.207, - "pct_cuda_time": 1.295659164611294, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.439, - "pct_cuda_time": 1.28981295242013, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 54.426, - "cuda_time_us": 22.752, - "pct_cuda_time": 0.17319403616323747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.752, - "pct_cuda_time": 0.17319403616323747, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.228, - "cuda_time_us": 90.655, - "pct_cuda_time": 0.6900890184765425, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 90.655, - "pct_cuda_time": 0.6900890184765425, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 946.636, - "cuda_time_us": 387.296, - "pct_cuda_time": 2.9481960895691466, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.771, - "cuda_time_us": 6.911, - "pct_cuda_time": 0.05260829746501996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.911, - "pct_cuda_time": 0.05260829746501996, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 684.69, - "cuda_time_us": 101.505, - "pct_cuda_time": 0.7726819901876504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.886, - "cuda_time_us": 44.416, - "pct_cuda_time": 0.33810593838899244, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.416, - "pct_cuda_time": 0.33810593838899244, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 198.153, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 306.655, - "cuda_time_us": 16.672, - "pct_cuda_time": 0.1269115229831881, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.02801310008266148, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.521, - "pct_cuda_time": 0.08770079512291927, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.471, - "pct_cuda_time": 0.011197627777607348, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.035, - "cuda_time_us": 33.665, - "pct_cuda_time": 0.25626657996815183, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.665, - "pct_cuda_time": 0.25626657996815183, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.347, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.987, - "cuda_time_us": 272.096, - "pct_cuda_time": 2.0712642608945266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.499, - "cuda_time_us": 169.37599999999998, - "pct_cuda_time": 1.2893333803263234, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.64, - "pct_cuda_time": 1.2837307603097912, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.238, - "cuda_time_us": 22.656, - "pct_cuda_time": 0.17246325963934198, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.656, - "pct_cuda_time": 0.17246325963934198, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.368, - "cuda_time_us": 80.064, - "pct_cuda_time": 0.609467620928861, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.064, - "pct_cuda_time": 0.609467620928861, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 925.68, - "cuda_time_us": 393.95300000000003, - "pct_cuda_time": 2.998870874148027, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.192, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 666.8, - "cuda_time_us": 102.209, - "pct_cuda_time": 0.778041018029551, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.214, - "cuda_time_us": 44.608, - "pct_cuda_time": 0.3395674914367835, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.608, - "pct_cuda_time": 0.3395674914367835, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 190.489, - "cuda_time_us": 7.328, - "pct_cuda_time": 0.055782607990691124, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.328, - "pct_cuda_time": 0.055782607990691124, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 296.46, - "cuda_time_us": 16.672, - "pct_cuda_time": 0.1269115229831881, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.028987468781188835, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.553, - "pct_cuda_time": 0.08794438729755112, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.311, - "pct_cuda_time": 0.009979666904448152, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 67.169, - "cuda_time_us": 33.601, - "pct_cuda_time": 0.2557793956188882, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.601, - "pct_cuda_time": 0.2557793956188882, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.298, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.816, - "pct_cuda_time": 0.0518851331965817, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.248, - "cuda_time_us": 278.016, - "pct_cuda_time": 2.116328813201417, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.948, - "cuda_time_us": 169.792, - "pct_cuda_time": 1.2925000785965375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.056, - "pct_cuda_time": 1.2868974585800053, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.712, - "cuda_time_us": 23.008, - "pct_cuda_time": 0.1751427735602922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 23.008, - "pct_cuda_time": 0.1751427735602922, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.009, - "cuda_time_us": 85.216, - "pct_cuda_time": 0.6486859610445871, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 85.216, - "pct_cuda_time": 0.6486859610445871, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 933.888, - "cuda_time_us": 387.037, - "pct_cuda_time": 2.9462245154057203, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.219, - "cuda_time_us": 7.199, - "pct_cuda_time": 0.05480062703670652, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.199, - "pct_cuda_time": 0.05480062703670652, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 662.953, - "cuda_time_us": 101.56700000000001, - "pct_cuda_time": 0.7731539500259996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.775, - "cuda_time_us": 45.343, - "pct_cuda_time": 0.3451624991978586, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.343, - "pct_cuda_time": 0.3451624991978586, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.811, - "cuda_time_us": 6.72, - "pct_cuda_time": 0.051154356672686176, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.72, - "pct_cuda_time": 0.051154356672686176, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 293.466, - "cuda_time_us": 16.736, - "pct_cuda_time": 0.1273987073324518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.028987468781188835, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.08720599851819834, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.011205240033064591, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 73.438, - "cuda_time_us": 32.768, - "pct_cuda_time": 0.2494383868230031, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 32.768, - "pct_cuda_time": 0.2494383868230031, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 35.757, - "cuda_time_us": 6.559, - "pct_cuda_time": 0.04992878354406974, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.559, - "pct_cuda_time": 0.04992878354406974, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.135, - "cuda_time_us": 271.712, - "pct_cuda_time": 2.0683411547989445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.832, - "cuda_time_us": 169.727, - "pct_cuda_time": 1.2920052819918166, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.005595007761075051, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.992, - "pct_cuda_time": 1.2864102742307415, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.072, - "cuda_time_us": 22.657, - "pct_cuda_time": 0.17247087189479923, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.657, - "pct_cuda_time": 0.17247087189479923, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.059, - "cuda_time_us": 79.328, - "pct_cuda_time": 0.6038650009123288, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.328, - "pct_cuda_time": 0.6038650009123288, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 924.484, - "cuda_time_us": 404.60900000000004, - "pct_cuda_time": 3.0799870683004293, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.824, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 658.235, - "cuda_time_us": 103.17000000000002, - "pct_cuda_time": 0.7853563955239633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.429, - "cuda_time_us": 45.409, - "pct_cuda_time": 0.3456649080580367, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.409, - "pct_cuda_time": 0.3456649080580367, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.655, - "cuda_time_us": 6.848, - "pct_cuda_time": 0.05212872537121353, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.848, - "pct_cuda_time": 0.05212872537121353, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 287.667, - "cuda_time_us": 16.673000000000002, - "pct_cuda_time": 0.12691913523864537, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.02801310008266148, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.489, - "pct_cuda_time": 0.08745720294828743, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.01144883220769643, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 70.156, - "cuda_time_us": 34.24, - "pct_cuda_time": 0.2606436268560677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 34.24, - "pct_cuda_time": 0.2606436268560677, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.86, - "cuda_time_us": 6.687, - "pct_cuda_time": 0.050903152242597095, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.687, - "pct_cuda_time": 0.050903152242597095, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.262, - "cuda_time_us": 288.096, - "pct_cuda_time": 2.1930603482104463, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.923, - "cuda_time_us": 169.92, - "pct_cuda_time": 1.2934744472950648, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.184, - "pct_cuda_time": 1.2878718272785326, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 41.026, - "cuda_time_us": 22.752, - "pct_cuda_time": 0.17319403616323747, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.752, - "pct_cuda_time": 0.17319403616323747, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.679, - "cuda_time_us": 95.424, - "pct_cuda_time": 0.7263918647521438, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 95.424, - "pct_cuda_time": 0.7263918647521438, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1047.88, - "cuda_time_us": 388.00300000000004, - "pct_cuda_time": 2.9535779541774194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.644, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.752, - "pct_cuda_time": 0.05139794884731802, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 787.101, - "cuda_time_us": 101.762, - "pct_cuda_time": 0.7746383398401623, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.545, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.3385931227382561, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.48, - "pct_cuda_time": 0.3385931227382561, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 200.008, - "cuda_time_us": 7.04, - "pct_cuda_time": 0.053590278419004565, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.04, - "pct_cuda_time": 0.053590278419004565, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 401.592, - "cuda_time_us": 16.641, - "pct_cuda_time": 0.1266755430640135, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.027525915733397802, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.552, - "pct_cuda_time": 0.08793677504209385, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.473, - "pct_cuda_time": 0.011212852288521837, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 74.142, - "cuda_time_us": 33.601, - "pct_cuda_time": 0.2557793956188882, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.601, - "pct_cuda_time": 0.2557793956188882, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.729, - "cuda_time_us": 6.689, - "pct_cuda_time": 0.05091837675351158, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.689, - "pct_cuda_time": 0.05091837675351158, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.172, - "cuda_time_us": 272.8, - "pct_cuda_time": 2.0766232887364273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.711, - "cuda_time_us": 169.536, - "pct_cuda_time": 1.2905513411994827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.768, - "pct_cuda_time": 1.2847051290083187, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.467, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.17416840486176483, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.17416840486176483, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.557, - "cuda_time_us": 80.384, - "pct_cuda_time": 0.6119035426751794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.384, - "pct_cuda_time": 0.6119035426751794, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 950.546, - "cuda_time_us": 387.679, - "pct_cuda_time": 2.9511115834092716, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.624, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 683.036, - "cuda_time_us": 102.33600000000001, - "pct_cuda_time": 0.779007774472621, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.538, - "cuda_time_us": 45.761, - "pct_cuda_time": 0.34834442197898696, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.761, - "pct_cuda_time": 0.34834442197898696, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 210.06, - "cuda_time_us": 6.432, - "pct_cuda_time": 0.04896202710099963, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.432, - "pct_cuda_time": 0.04896202710099963, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 296.315, - "cuda_time_us": 16.576, - "pct_cuda_time": 0.1261807464592926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.02825669225729332, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.552, - "pct_cuda_time": 0.08793677504209385, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.009987279159905398, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.278, - "cuda_time_us": 33.567, - "pct_cuda_time": 0.2555205789333418, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.567, - "pct_cuda_time": 0.2555205789333418, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 37.449, - "cuda_time_us": 6.719, - "pct_cuda_time": 0.05114674441722894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.719, - "pct_cuda_time": 0.05114674441722894, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 174.064, - "cuda_time_us": 271.64799999999997, - "pct_cuda_time": 2.0678539704496806, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.144, - "cuda_time_us": 169.504, - "pct_cuda_time": 1.290307749024851, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.768, - "pct_cuda_time": 1.2847051290083187, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.183, - "cuda_time_us": 22.848, - "pct_cuda_time": 0.173924812687133, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.848, - "pct_cuda_time": 0.173924812687133, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.534, - "cuda_time_us": 79.296, - "pct_cuda_time": 0.603621408737697, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.296, - "pct_cuda_time": 0.603621408737697, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 891.41, - "cuda_time_us": 388.48, - "pct_cuda_time": 2.957209000030525, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.515, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.88, - "pct_cuda_time": 0.05237231754584538, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 632.912, - "cuda_time_us": 102.56, - "pct_cuda_time": 0.7807129196950439, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.743, - "cuda_time_us": 45.664, - "pct_cuda_time": 0.3476060331996342, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.664, - "pct_cuda_time": 0.3476060331996342, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 180.694, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 280.771, - "cuda_time_us": 16.544, - "pct_cuda_time": 0.12593715428466076, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.028743876606556997, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.08720599851819834, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.009987279159905398, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 67.496, - "cuda_time_us": 33.44, - "pct_cuda_time": 0.2545538224902717, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.44, - "pct_cuda_time": 0.2545538224902717, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.137, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.784, - "pct_cuda_time": 0.05164154102194986, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.277, - "cuda_time_us": 272.25600000000003, - "pct_cuda_time": 2.072482221767686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.019, - "cuda_time_us": 169.6, - "pct_cuda_time": 1.2910385255487464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.005610232271989541, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.863, - "pct_cuda_time": 1.285428293276757, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.881, - "cuda_time_us": 23.616, - "pct_cuda_time": 0.17977102487829716, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 23.616, - "pct_cuda_time": 0.17977102487829716, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.787, - "cuda_time_us": 79.04, - "pct_cuda_time": 0.6016726713406423, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.04, - "pct_cuda_time": 0.6016726713406423, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 954.004, - "cuda_time_us": 401.824, - "pct_cuda_time": 3.0587869368520018, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.25, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.912, - "pct_cuda_time": 0.052615909720477215, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 698.312, - "cuda_time_us": 103.07300000000001, - "pct_cuda_time": 0.7846180067446106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.176, - "cuda_time_us": 45.856, - "pct_cuda_time": 0.34906758624742523, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 45.856, - "pct_cuda_time": 0.34906758624742523, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 201.539, - "cuda_time_us": 7.264, - "pct_cuda_time": 0.05529542364142745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.264, - "pct_cuda_time": 0.05529542364142745, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 320.093, - "cuda_time_us": 16.641000000000002, - "pct_cuda_time": 0.12667554306401352, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.713, - "pct_cuda_time": 0.028264304512750565, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.456, - "pct_cuda_time": 0.08720599851819834, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.011205240033064591, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 70.127, - "cuda_time_us": 33.312, - "pct_cuda_time": 0.25357945379174435, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.312, - "pct_cuda_time": 0.25357945379174435, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.934, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.688, - "pct_cuda_time": 0.050910764498054345, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.359, - "cuda_time_us": 285.151, - "pct_cuda_time": 2.17064225588886, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.929, - "cuda_time_us": 169.471, - "pct_cuda_time": 1.2900565445947618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 168.703, - "pct_cuda_time": 1.2842103324035978, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.21, - "cuda_time_us": 22.848, - "pct_cuda_time": 0.173924812687133, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.848, - "pct_cuda_time": 0.173924812687133, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.733, - "cuda_time_us": 92.832, - "pct_cuda_time": 0.7066608986069647, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 92.832, - "pct_cuda_time": 0.7066608986069647, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 884.703, - "cuda_time_us": 404.481, - "pct_cuda_time": 3.0790126996019014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.038, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 7.008, - "pct_cuda_time": 0.053346686244372735, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 630.566, - "cuda_time_us": 102.305, - "pct_cuda_time": 0.7787717945534465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 49.643, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.34175982100847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.34175982100847, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.602, - "cuda_time_us": 6.592, - "pct_cuda_time": 0.05017998797415882, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.592, - "pct_cuda_time": 0.05017998797415882, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 272.439, - "cuda_time_us": 16.673000000000002, - "pct_cuda_time": 0.12691913523864537, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.02801310008266148, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.521, - "pct_cuda_time": 0.08770079512291927, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.011205240033064591, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.842, - "cuda_time_us": 34.144, - "pct_cuda_time": 0.25991285033217215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 34.144, - "pct_cuda_time": 0.25991285033217215, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.545, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.656, - "pct_cuda_time": 0.0506671723234225, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.53, - "cuda_time_us": 288.512, - "pct_cuda_time": 2.1962270464806597, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.171, - "cuda_time_us": 170.272, - "pct_cuda_time": 1.2961539612160151, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.536, - "pct_cuda_time": 1.2905513411994827, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.983, - "cuda_time_us": 22.784, - "pct_cuda_time": 0.17343762833786933, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.784, - "pct_cuda_time": 0.17343762833786933, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.062, - "cuda_time_us": 95.456, - "pct_cuda_time": 0.7266354569267757, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 95.456, - "pct_cuda_time": 0.7266354569267757, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 951.596, - "cuda_time_us": 398.046, - "pct_cuda_time": 3.03002783573453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.954, - "cuda_time_us": 6.879, - "pct_cuda_time": 0.05236470529038813, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.879, - "pct_cuda_time": 0.05236470529038813, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 662.684, - "cuda_time_us": 101.63199999999999, - "pct_cuda_time": 0.7736487466307204, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.951, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.33567001664267404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.096, - "pct_cuda_time": 0.33567001664267404, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[512, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 193.18, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "_C::rotary_embedding(int64[512], bfloat16[512, 4096], bfloat16[512, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 290.719, - "cuda_time_us": 16.736, - "pct_cuda_time": 0.1273987073324518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.028500284431925156, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[512], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 11.68, - "pct_cuda_time": 0.08891114374062122, - "trace": "_vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.009987279159905398, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], None, None, bfloat16[512, 32, 128], int32[3], int32[3], None, None, None, 256, 256, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[512, 32, 128], bfloat16[512, 8, 128], bfloat16[512, 8, 128], bfloat16[512, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.514, - "cuda_time_us": 33.824, - "pct_cuda_time": 0.25747692858585375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 33.824, - "pct_cuda_time": 0.25747692858585375, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[512, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 47.483, - "cuda_time_us": 6.592, - "pct_cuda_time": 0.05017998797415882, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.592, - "pct_cuda_time": 0.05017998797415882, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 183.129, - "cuda_time_us": 282.943, - "pct_cuda_time": 2.1538343958392625, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.839, - "cuda_time_us": 169.984, - "pct_cuda_time": 1.2939616316443285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 169.216, - "pct_cuda_time": 1.2881154194531645, - "trace": "mm(bfloat16[512, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[512, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[512, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 42.174, - "cuda_time_us": 22.751, - "pct_cuda_time": 0.17318642390778025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 22.751, - "pct_cuda_time": 0.17318642390778025, - "trace": "_C::silu_and_mul(bfloat16[512, 14336], bfloat16[512, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.95, - "cuda_time_us": 90.208, - "pct_cuda_time": 0.686686340287154, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 90.208, - "pct_cuda_time": 0.686686340287154, - "trace": "mm(bfloat16[512, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[512, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[512, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.313, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 6.976, - "pct_cuda_time": 0.05310309406974089, - "trace": "_C::fused_add_rms_norm(bfloat16[512, 4096], bfloat16[512, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 156.665, - "cuda_time_us": 354.71999999999997, - "pct_cuda_time": 2.7002192557939346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.072, - "pct_cuda_time": 0.02338484876465654, - "trace": "index_select(bfloat16[512, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0056026200165322955, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 350.912, - "pct_cuda_time": 2.6712317870127458, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 1189.552, - "cuda_time_us": 118.817, - "pct_cuda_time": 0.9044653566634753, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.177, - "pct_cuda_time": 0.016571880130422295, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.01680786004959689, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.239, - "pct_cuda_time": 0.01704383996877148, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.01656426787496505, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.005846212191164135, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.8, - "pct_cuda_time": 0.006089804365795974, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.769, - "pct_cuda_time": 0.00585382444662138, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.096, - "pct_cuda_time": 0.031179798352875387, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.64, - "pct_cuda_time": 0.03532086532161665, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.016, - "pct_cuda_time": 0.2589384816336448, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.776, - "pct_cuda_time": 0.21143800758043624, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 2.048, - "pct_cuda_time": 0.015589899176437693, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.832, - "pct_cuda_time": 0.03678241836940768, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 28.0, - "pct_cuda_time": 0.2131431528028591, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.017295044398860565, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - }, - "decode_1": { - "metadata": { - "num_running_seqs": 2 - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 6336.215000000002, - "pct_cuda_time": 93.11051149088149, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 3.68, - "pct_cuda_time": 0.054077502465816545, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.68, - "pct_cuda_time": 0.054077502465816545, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 6329.334000000002, - "pct_cuda_time": 93.0093953782545, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 205.919, - "pct_cuda_time": 3.025974247352847, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.256, - "pct_cuda_time": 0.06254180719959652, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 201.663, - "pct_cuda_time": 2.9634324401532504, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 1866.7430000000002, - "pct_cuda_time": 27.4317389091157, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 670.593, - "pct_cuda_time": 9.8543463617009, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 670.593, - "pct_cuda_time": 9.8543463617009, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 116.54399999999998, - "pct_cuda_time": 1.7126109911348157, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 116.54399999999998, - "pct_cuda_time": 1.7126109911348157, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 517.497, - "pct_cuda_time": 7.604604699334962, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 73.66, - "pct_cuda_time": 1.0824317477260994, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 401.118, - "pct_cuda_time": 5.894418378827011, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 42.719, - "pct_cuda_time": 0.6277545727818524, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 562.109, - "pct_cuda_time": 8.26017685694502, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cuda_time_us": 492.63599999999985, - "pct_cuda_time": 7.23927296324728, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cuda_time_us": 69.47300000000001, - "pct_cuda_time": 1.0209038936977373, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 4256.672000000001, - "pct_cuda_time": 62.55168222178595, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 2583.168, - "pct_cuda_time": 37.95958529609195, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2583.168, - "pct_cuda_time": 37.95958529609195, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 284.96399999999994, - "pct_cuda_time": 4.18753842735569, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 284.96399999999994, - "pct_cuda_time": 4.18753842735569, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 1388.54, - "pct_cuda_time": 20.40455849833829, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1388.54, - "pct_cuda_time": 20.40455849833829, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 3.201, - "pct_cuda_time": 0.0470386101611627, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 3.201, - "pct_cuda_time": 0.0470386101611627, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 347.265, - "pct_cuda_time": 5.10304997113907, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.041, - "pct_cuda_time": 0.044687414401779375, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010815500493163308, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 343.488, - "pct_cuda_time": 5.047547056244128, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 121.568, - "pct_cuda_time": 1.7864385379794525, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 12.703000000000001, - "pct_cuda_time": 0.1866702483215401, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.128, - "pct_cuda_time": 0.06066085059208986, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.8, - "pct_cuda_time": 0.07053587278149982, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 34.432, - "pct_cuda_time": 0.5059773274192922, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 28.032, - "pct_cuda_time": 0.41192949704395904, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 1.984, - "pct_cuda_time": 0.029154827416353262, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.928, - "pct_cuda_time": 0.0724168293890065, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 28.064, - "pct_cuda_time": 0.41239973619583575, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.497, - "pct_cuda_time": 0.03669334881987606, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 30022.089, - "cuda_time_us": 6336.215000000002, - "pct_cuda_time": 93.11051149088149, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 87.055, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.054077502465816545, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.054077502465816545, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[2]) <- embedding(bfloat16[128256, 4096], int64[2], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1360.681, - "cuda_time_us": 203.36200000000002, - "pct_cuda_time": 2.9883992001232023, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 76.547, - "cuda_time_us": 4.256, - "pct_cuda_time": 0.06254180719959652, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.256, - "pct_cuda_time": 0.06254180719959652, - "trace": "_C::rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 978.628, - "cuda_time_us": 62.049, - "pct_cuda_time": 0.9118084104623506, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 126.034, - "cuda_time_us": 24.0, - "pct_cuda_time": 0.35267936390749915, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 24.0, - "pct_cuda_time": 0.35267936390749915, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 327.146, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 347.85, - "cuda_time_us": 16.385, - "pct_cuda_time": 0.24077714073434892, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.368, - "pct_cuda_time": 0.03479769723887325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.704, - "pct_cuda_time": 0.18668494329503624, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.313, - "pct_cuda_time": 0.01929450020043943, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 90.959, - "cuda_time_us": 18.08, - "pct_cuda_time": 0.26568512081031603, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.744, - "pct_cuda_time": 0.23135766272331945, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03432745808699659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 40.577, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045613197732036564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045613197732036564, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 223.541, - "cuda_time_us": 133.953, - "pct_cuda_time": 1.9684357847292184, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 81.88, - "cuda_time_us": 81.536, - "pct_cuda_time": 1.198169358981744, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.536, - "pct_cuda_time": 1.198169358981744, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 50.434, - "cuda_time_us": 8.897, - "pct_cuda_time": 0.1307411791952092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.897, - "pct_cuda_time": 0.1307411791952092, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 63.786, - "cuda_time_us": 43.52, - "pct_cuda_time": 0.6395252465522652, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.52, - "pct_cuda_time": 0.6395252465522652, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 971.259, - "cuda_time_us": 198.495, - "pct_cuda_time": 2.9168787641174605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 40.421, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 686.732, - "cuda_time_us": 58.305, - "pct_cuda_time": 0.8567904296927809, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.626, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.3042447312642026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.3042447312642026, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 211.51, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 291.411, - "cuda_time_us": 16.257, - "pct_cuda_time": 0.23889618412684227, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.305, - "pct_cuda_time": 0.03387191390861607, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.18527422583940623, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.337, - "cuda_time_us": 17.759999999999998, - "pct_cuda_time": 0.26098272929154936, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.648, - "pct_cuda_time": 0.22994694526768944, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.031035784023859928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.359, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.88, - "cuda_time_us": 133.854, - "pct_cuda_time": 1.9669809823531, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.089, - "cuda_time_us": 81.983, - "pct_cuda_time": 1.2047380121345213, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.983, - "pct_cuda_time": 1.2047380121345213, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.981, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.13072648422171304, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.13072648422171304, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.12, - "cuda_time_us": 42.975, - "pct_cuda_time": 0.6315164859968657, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.975, - "pct_cuda_time": 0.6315164859968657, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 893.875, - "cuda_time_us": 197.40800000000002, - "pct_cuda_time": 2.90090532792715, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.978, - "cuda_time_us": 3.231, - "pct_cuda_time": 0.04747945936604707, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.231, - "pct_cuda_time": 0.04747945936604707, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 626.97, - "cuda_time_us": 57.888, - "pct_cuda_time": 0.8506626257448879, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.972, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.3042447312642026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.3042447312642026, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 188.596, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 269.162, - "cuda_time_us": 16.096, - "pct_cuda_time": 0.2365302933939628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.271, - "pct_cuda_time": 0.03337228480974711, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.48, - "pct_cuda_time": 0.18339326923189958, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.345, - "pct_cuda_time": 0.0197647393523161, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.14, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.2567505769246594, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.2257147929007995, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.031035784023859928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.18, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.275, - "cuda_time_us": 133.089, - "pct_cuda_time": 1.9557393276285484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.602, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.182181227817937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.182181227817937, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.262, - "cuda_time_us": 8.993, - "pct_cuda_time": 0.13215189665083918, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.993, - "pct_cuda_time": 0.13215189665083918, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.496, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6414062031597719, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6414062031597719, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 873.122, - "cuda_time_us": 197.984, - "pct_cuda_time": 2.90936963266093, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.784, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 613.383, - "cuda_time_us": 58.112, - "pct_cuda_time": 0.8539542998080247, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.581, - "cuda_time_us": 20.577, - "pct_cuda_time": 0.30237846963019216, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.577, - "pct_cuda_time": 0.30237846963019216, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 187.556, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 261.841, - "cuda_time_us": 16.447000000000003, - "pct_cuda_time": 0.24168822909110999, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.368, - "pct_cuda_time": 0.03479769723887325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.736, - "pct_cuda_time": 0.1871551824469129, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.019735349405323805, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.279, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.2567505769246594, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.2252445537489228, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.59, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 174.918, - "cuda_time_us": 133.536, - "pct_cuda_time": 1.9623079807813255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.985, - "cuda_time_us": 80.928, - "pct_cuda_time": 1.1892348150960872, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.928, - "pct_cuda_time": 1.1892348150960872, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.525, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.13025624506983638, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.13025624506983638, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.422, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.6428169206154019, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.744, - "pct_cuda_time": 0.6428169206154019, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 903.58, - "cuda_time_us": 197.981, - "pct_cuda_time": 2.9093255477404414, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.794, - "cuda_time_us": 3.167, - "pct_cuda_time": 0.04653898106229374, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.167, - "pct_cuda_time": 0.04653898106229374, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 644.771, - "cuda_time_us": 58.846000000000004, - "pct_cuda_time": 0.8647404103541957, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.898, - "cuda_time_us": 20.863, - "pct_cuda_time": 0.3065812320500898, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.863, - "pct_cuda_time": 0.3065812320500898, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 208.25, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 269.231, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.23841125000146945, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03432745808699659, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.576, - "pct_cuda_time": 0.1848039866875296, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.79, - "cuda_time_us": 18.143, - "pct_cuda_time": 0.26661090414057326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.032, - "pct_cuda_time": 0.23558981509020946, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.031021089050363784, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.547, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.428, - "cuda_time_us": 132.768, - "pct_cuda_time": 1.9510222411362854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.115, - "cuda_time_us": 80.608, - "pct_cuda_time": 1.1845324235773205, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.608, - "pct_cuda_time": 1.1845324235773205, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.09, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.13683959319610967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.13683959319610967, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.261, - "cuda_time_us": 42.848, - "pct_cuda_time": 0.6296502243628552, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.848, - "pct_cuda_time": 0.6296502243628552, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 976.989, - "cuda_time_us": 197.023, - "pct_cuda_time": 2.8952477631311337, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.621, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 679.872, - "cuda_time_us": 57.951, - "pct_cuda_time": 0.8515884090751453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.959, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.305185209567956, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.305185209567956, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 182.391, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 327.445, - "cuda_time_us": 16.191, - "pct_cuda_time": 0.2379263158760966, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.575, - "pct_cuda_time": 0.1847892917140334, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.892, - "cuda_time_us": 17.344, - "pct_cuda_time": 0.2548696203171527, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.2, - "pct_cuda_time": 0.22336359714141613, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.527, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 182.968, - "cuda_time_us": 132.768, - "pct_cuda_time": 1.9510222411362854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.81, - "cuda_time_us": 81.056, - "pct_cuda_time": 1.1911157717035938, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.056, - "pct_cuda_time": 1.1911157717035938, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.524, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.736, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6315311809703618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6315311809703618, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 897.898, - "cuda_time_us": 197.44, - "pct_cuda_time": 2.9013755670790267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.772, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 647.159, - "cuda_time_us": 57.92, - "pct_cuda_time": 0.8511328648967648, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.541, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.3047149704160793, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.3047149704160793, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 197.544, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 272.844, - "cuda_time_us": 16.0, - "pct_cuda_time": 0.2351195759383328, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.416, - "pct_cuda_time": 0.18245279092814626, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.655, - "cuda_time_us": 17.6, - "pct_cuda_time": 0.25863153353216606, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.487, - "pct_cuda_time": 0.22758105453481, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.031050478997356072, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.496, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.152, - "cuda_time_us": 133.12, - "pct_cuda_time": 1.9561948718069286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.181, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.182181227817937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.182181227817937, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.913, - "cuda_time_us": 9.376, - "pct_cuda_time": 0.137780071499863, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.376, - "pct_cuda_time": 0.137780071499863, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.92, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.6362335724891285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.6362335724891285, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 846.11, - "cuda_time_us": 197.696, - "pct_cuda_time": 2.90513748029404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.802, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 602.448, - "cuda_time_us": 58.048, - "pct_cuda_time": 0.8530138215042714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.508, - "cuda_time_us": 20.832, - "pct_cuda_time": 0.3061256878717093, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.832, - "pct_cuda_time": 0.3061256878717093, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 178.619, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 262.85, - "cuda_time_us": 16.256, - "pct_cuda_time": 0.23888148915334612, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.271, - "pct_cuda_time": 0.03337228480974711, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.641, - "pct_cuda_time": 0.18575915996477904, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.042, - "cuda_time_us": 17.311999999999998, - "pct_cuda_time": 0.25439938116527605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.168, - "pct_cuda_time": 0.22289335798953946, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.728, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 165.629, - "cuda_time_us": 133.344, - "pct_cuda_time": 1.9594865458700654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 57.044, - "cuda_time_us": 80.192, - "pct_cuda_time": 1.1784193146029238, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.192, - "pct_cuda_time": 1.1784193146029238, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.539, - "cuda_time_us": 8.48, - "pct_cuda_time": 0.12461337524731639, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.48, - "pct_cuda_time": 0.12461337524731639, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.832, - "cuda_time_us": 44.672, - "pct_cuda_time": 0.6564538560198251, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.672, - "pct_cuda_time": 0.6564538560198251, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 861.224, - "cuda_time_us": 199.039, - "pct_cuda_time": 2.9248728296993636, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.285, - "cuda_time_us": 3.295, - "pct_cuda_time": 0.048419937669800406, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.295, - "pct_cuda_time": 0.048419937669800406, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 603.246, - "cuda_time_us": 58.879999999999995, - "pct_cuda_time": 0.8652400394530645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.081, - "cuda_time_us": 21.503, - "pct_cuda_time": 0.31598601508762314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.503, - "pct_cuda_time": 0.31598601508762314, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 178.781, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 261.453, - "cuda_time_us": 16.096, - "pct_cuda_time": 0.2365302933939628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.481, - "pct_cuda_time": 0.1834079642053957, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.019735349405323805, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.587, - "cuda_time_us": 17.697, - "pct_cuda_time": 0.2600569459612922, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.552, - "pct_cuda_time": 0.22853622781205946, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.145, - "pct_cuda_time": 0.03152071814923274, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.127, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.392, - "cuda_time_us": 133.728, - "pct_cuda_time": 1.9651294156925854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.399, - "cuda_time_us": 81.536, - "pct_cuda_time": 1.198169358981744, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.536, - "pct_cuda_time": 1.198169358981744, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.99, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12978600591795972, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12978600591795972, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.449, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.6371740507928819, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.6371740507928819, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 921.96, - "cuda_time_us": 197.565, - "pct_cuda_time": 2.9032124387660447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.225, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047964393491419885, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047964393491419885, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 667.646, - "cuda_time_us": 58.11, - "pct_cuda_time": 0.8539249098610323, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 84.834, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.30377449211232593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.30377449211232593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 198.838, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 261.117, - "cuda_time_us": 16.415, - "pct_cuda_time": 0.24121798993923327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03667865384637991, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.18527422583940623, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.311, - "pct_cuda_time": 0.019265110253447144, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 67.228, - "cuda_time_us": 17.407, - "pct_cuda_time": 0.25579540364740994, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.263, - "pct_cuda_time": 0.2242893804716733, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.34, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.04936041597355374, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.359, - "pct_cuda_time": 0.04936041597355374, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.786, - "cuda_time_us": 132.832, - "pct_cuda_time": 1.9519627194400386, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.443, - "cuda_time_us": 80.704, - "pct_cuda_time": 1.1859431410329504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.704, - "pct_cuda_time": 1.1859431410329504, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.494, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1288455276142064, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1288455276142064, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.897, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.6371740507928819, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.36, - "pct_cuda_time": 0.6371740507928819, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 883.697, - "cuda_time_us": 198.879, - "pct_cuda_time": 2.92252163393998, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.452, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045613197732036564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045613197732036564, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 613.932, - "cuda_time_us": 58.079, - "pct_cuda_time": 0.8534693656826517, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.46, - "cuda_time_us": 21.088, - "pct_cuda_time": 0.30988760108672264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.088, - "pct_cuda_time": 0.30988760108672264, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 182.057, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 261.393, - "cuda_time_us": 16.096, - "pct_cuda_time": 0.2365302933939628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03432745808699659, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.18292303008002292, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 64.026, - "cuda_time_us": 17.311, - "pct_cuda_time": 0.2543846861917799, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.167, - "pct_cuda_time": 0.22287866301604334, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.8, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 184.913, - "cuda_time_us": 134.528, - "pct_cuda_time": 1.976885394489502, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 70.705, - "cuda_time_us": 81.248, - "pct_cuda_time": 1.193937206614854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.248, - "pct_cuda_time": 1.193937206614854, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.462, - "cuda_time_us": 8.96, - "pct_cuda_time": 0.13166696252546636, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.96, - "pct_cuda_time": 0.13166696252546636, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.802, - "cuda_time_us": 44.32, - "pct_cuda_time": 0.6512812253491819, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.32, - "pct_cuda_time": 0.6512812253491819, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 930.808, - "cuda_time_us": 196.54399999999998, - "pct_cuda_time": 2.88820887082648, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.872, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 663.687, - "cuda_time_us": 58.206999999999994, - "pct_cuda_time": 0.8553503222901585, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.923, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.30377449211232593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.30377449211232593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 193.631, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05642869822519987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.84, - "pct_cuda_time": 0.05642869822519987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 262.725, - "cuda_time_us": 15.904, - "pct_cuda_time": 0.23370885848270276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03291674063136659, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.32, - "pct_cuda_time": 0.18104207347251625, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 61.263, - "cuda_time_us": 17.791, - "pct_cuda_time": 0.26143827346992987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.648, - "pct_cuda_time": 0.22994694526768944, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.031491328202240446, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.496, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.04609813185740937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.04609813185740937, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.475, - "cuda_time_us": 132.0, - "pct_cuda_time": 1.9397365014912453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.856, - "cuda_time_us": 79.84, - "pct_cuda_time": 1.1732466839322806, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.84, - "pct_cuda_time": 1.1732466839322806, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 45.92, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.13307767998109635, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.056, - "pct_cuda_time": 0.13307767998109635, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.376, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.6334121375778685, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.6334121375778685, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 951.216, - "cuda_time_us": 198.62599999999998, - "pct_cuda_time": 2.9188038056454553, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.058, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 680.665, - "cuda_time_us": 58.209, - "pct_cuda_time": 0.8553797122371509, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.194, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.30659592702358596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.30659592702358596, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.134, - "cuda_time_us": 3.777, - "pct_cuda_time": 0.05550291489494269, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.777, - "pct_cuda_time": 0.05550291489494269, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 332.861, - "cuda_time_us": 16.256, - "pct_cuda_time": 0.23888148915334612, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.64, - "pct_cuda_time": 0.18574446499128291, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.899, - "cuda_time_us": 17.311999999999998, - "pct_cuda_time": 0.25439938116527605, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.2, - "pct_cuda_time": 0.22336359714141613, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.031035784023859928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.669, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.04750884931303937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.04750884931303937, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 187.962, - "cuda_time_us": 134.016, - "pct_cuda_time": 1.9693615680594752, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.322, - "cuda_time_us": 81.248, - "pct_cuda_time": 1.193937206614854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.248, - "pct_cuda_time": 1.193937206614854, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.92, - "cuda_time_us": 8.96, - "pct_cuda_time": 0.13166696252546636, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.96, - "pct_cuda_time": 0.13166696252546636, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 62.241, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6437573989191552, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6437573989191552, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 901.559, - "cuda_time_us": 197.312, - "pct_cuda_time": 2.89949461047152, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.966, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047964393491419885, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047964393491419885, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 637.828, - "cuda_time_us": 58.272, - "pct_cuda_time": 0.856305495567408, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.315, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.30659592702358596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.864, - "pct_cuda_time": 0.30659592702358596, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 193.126, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 273.931, - "cuda_time_us": 16.416, - "pct_cuda_time": 0.24123268491272942, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03291674063136659, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.832, - "pct_cuda_time": 0.1885658999025429, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.277, - "cuda_time_us": 17.375999999999998, - "pct_cuda_time": 0.2553398594690294, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.232, - "pct_cuda_time": 0.2238338362932928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.165, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.008, - "cuda_time_us": 132.608, - "pct_cuda_time": 1.9486710453769023, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.011, - "cuda_time_us": 80.608, - "pct_cuda_time": 1.1845324235773205, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.608, - "pct_cuda_time": 1.1845324235773205, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.648, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1288455276142064, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.1288455276142064, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.721, - "cuda_time_us": 43.232, - "pct_cuda_time": 0.6352930941853752, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.232, - "pct_cuda_time": 0.6352930941853752, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 897.949, - "cuda_time_us": 197.601, - "pct_cuda_time": 2.903741457811906, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 35.631, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 634.429, - "cuda_time_us": 58.529, - "pct_cuda_time": 0.8600821037559174, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 66.435, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.30377449211232593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.30377449211232593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.441, - "cuda_time_us": 3.585, - "pct_cuda_time": 0.05268147998368269, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.585, - "pct_cuda_time": 0.05268147998368269, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 270.931, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.23841125000146945, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03385721893511992, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.18527422583940623, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.985, - "cuda_time_us": 18.048000000000002, - "pct_cuda_time": 0.26521488165843937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.904, - "pct_cuda_time": 0.23370885848270276, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.534, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.32, - "cuda_time_us": 132.704, - "pct_cuda_time": 1.9500817628325322, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.689, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.177949075451047, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.177949075451047, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.122, - "cuda_time_us": 9.248, - "pct_cuda_time": 0.13589911489235634, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.248, - "pct_cuda_time": 0.13589911489235634, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.759, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.6362335724891285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.6362335724891285, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 874.555, - "cuda_time_us": 196.416, - "pct_cuda_time": 2.886327914218973, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.572, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047964393491419885, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047964393491419885, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 612.712, - "cuda_time_us": 57.952, - "pct_cuda_time": 0.8516031040486414, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.376, - "cuda_time_us": 21.025, - "pct_cuda_time": 0.3089618177564654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.025, - "pct_cuda_time": 0.3089618177564654, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 175.299, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.055488219921446535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.055488219921446535, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 263.143, - "cuda_time_us": 15.968, - "pct_cuda_time": 0.2346493367864561, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.352, - "pct_cuda_time": 0.1815123126243929, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 62.36, - "cuda_time_us": 17.183, - "pct_cuda_time": 0.25250372958427325, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.072, - "pct_cuda_time": 0.22148264053390948, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.031021089050363784, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.717, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.548, - "cuda_time_us": 132.064, - "pct_cuda_time": 1.9406769797949988, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.267, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.182181227817937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.182181227817937, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.338, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.325, - "cuda_time_us": 42.88, - "pct_cuda_time": 0.6301204635147318, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.88, - "pct_cuda_time": 0.6301204635147318, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 907.447, - "cuda_time_us": 197.376, - "pct_cuda_time": 2.9004350887752732, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.126, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045613197732036564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.045613197732036564, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 649.716, - "cuda_time_us": 58.176, - "pct_cuda_time": 0.854894778111778, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.814, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.30565544871983263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.30565544871983263, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 195.948, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 267.999, - "cuda_time_us": 16.192, - "pct_cuda_time": 0.23794101084959277, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03385721893511992, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.416, - "pct_cuda_time": 0.18245279092814626, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021631000986326615, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 62.252, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2576910552284128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.22665527120455278, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.031035784023859928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.566, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.144, - "cuda_time_us": 132.928, - "pct_cuda_time": 1.9533734368956688, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.604, - "cuda_time_us": 81.632, - "pct_cuda_time": 1.1995800764373739, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.632, - "pct_cuda_time": 1.1995800764373739, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.828, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12790504931045305, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12790504931045305, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.467, - "cuda_time_us": 42.592, - "pct_cuda_time": 0.6258883111478418, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.592, - "pct_cuda_time": 0.6258883111478418, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 846.791, - "cuda_time_us": 196.382, - "pct_cuda_time": 2.8858282851201045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.575, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 601.267, - "cuda_time_us": 57.919, - "pct_cuda_time": 0.8511181699232685, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.43, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.3014232963529426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.512, - "pct_cuda_time": 0.3014232963529426, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.295, - "cuda_time_us": 3.551, - "pct_cuda_time": 0.05218185088481373, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.551, - "pct_cuda_time": 0.05218185088481373, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 253.282, - "cuda_time_us": 16.0, - "pct_cuda_time": 0.2351195759383328, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03291674063136659, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.416, - "pct_cuda_time": 0.18245279092814626, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 57.624, - "cuda_time_us": 17.855999999999998, - "pct_cuda_time": 0.26239344674717935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.168, - "pct_cuda_time": 0.22289335798953946, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.688, - "pct_cuda_time": 0.03950008875763991, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.664, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 166.309, - "cuda_time_us": 132.063, - "pct_cuda_time": 1.9406622848215025, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.545, - "cuda_time_us": 79.52, - "pct_cuda_time": 1.1685442924135139, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.52, - "pct_cuda_time": 1.1685442924135139, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.915, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.061, - "cuda_time_us": 43.807, - "pct_cuda_time": 0.643742703945659, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.807, - "pct_cuda_time": 0.643742703945659, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 906.112, - "cuda_time_us": 198.56099999999998, - "pct_cuda_time": 2.9178486323682056, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.111, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 638.228, - "cuda_time_us": 58.048, - "pct_cuda_time": 0.8530138215042714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.018, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30283401380857267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30283401380857267, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 187.563, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 281.653, - "cuda_time_us": 16.224, - "pct_cuda_time": 0.23841125000146945, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03385721893511992, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.18527422583940623, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 64.378, - "cuda_time_us": 17.567999999999998, - "pct_cuda_time": 0.25816129438028934, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.392, - "pct_cuda_time": 0.22618503205267612, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03197626232761326, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.039, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 182.923, - "cuda_time_us": 134.177, - "pct_cuda_time": 1.9717274587923546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.323, - "cuda_time_us": 81.953, - "pct_cuda_time": 1.2042971629296366, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.953, - "pct_cuda_time": 1.2042971629296366, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.908, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.12602409270294637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.576, - "pct_cuda_time": 0.12602409270294637, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 57.144, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6414062031597719, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.648, - "pct_cuda_time": 0.6414062031597719, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 979.651, - "cuda_time_us": 196.67200000000003, - "pct_cuda_time": 2.890089827433987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.8, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 641.628, - "cuda_time_us": 57.696, - "pct_cuda_time": 0.847841190833628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.726, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30283401380857267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.30283401380857267, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 174.983, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 298.804, - "cuda_time_us": 15.936, - "pct_cuda_time": 0.23417909763457945, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.32, - "pct_cuda_time": 0.18104207347251625, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 61.078, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2576910552284128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.22665527120455278, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.031035784023859928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.057, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 254.326, - "cuda_time_us": 132.704, - "pct_cuda_time": 1.9500817628325322, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.622, - "cuda_time_us": 80.607, - "pct_cuda_time": 1.1845177286038244, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.607, - "pct_cuda_time": 1.1845177286038244, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.474, - "cuda_time_us": 8.833, - "pct_cuda_time": 0.12980070089145584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.833, - "pct_cuda_time": 0.12980070089145584, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 134.517, - "cuda_time_us": 43.264, - "pct_cuda_time": 0.6357633333372519, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.264, - "pct_cuda_time": 0.6357633333372519, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 872.268, - "cuda_time_us": 198.144, - "pct_cuda_time": 2.911720828420313, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.167, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 621.836, - "cuda_time_us": 58.398999999999994, - "pct_cuda_time": 0.8581717572014184, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.494, - "cuda_time_us": 20.96, - "pct_cuda_time": 0.30800664447921594, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.96, - "pct_cuda_time": 0.30800664447921594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 181.758, - "cuda_time_us": 3.583, - "pct_cuda_time": 0.0526520900366904, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.583, - "pct_cuda_time": 0.0526520900366904, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 274.509, - "cuda_time_us": 16.352, - "pct_cuda_time": 0.2402922066089761, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.03526793639074991, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.64, - "pct_cuda_time": 0.18574446499128291, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.494, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.2572208160765361, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.264, - "pct_cuda_time": 0.22430407544516948, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03291674063136659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.143, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.789, - "cuda_time_us": 133.409, - "pct_cuda_time": 1.960441719147315, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.227, - "cuda_time_us": 80.32, - "pct_cuda_time": 1.1803002712104305, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.32, - "pct_cuda_time": 1.1803002712104305, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.504, - "cuda_time_us": 9.153, - "pct_cuda_time": 0.13450309241022249, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.153, - "pct_cuda_time": 0.13450309241022249, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.265, - "cuda_time_us": 43.936, - "pct_cuda_time": 0.6456383555266618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.936, - "pct_cuda_time": 0.6456383555266618, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 892.619, - "cuda_time_us": 196.381, - "pct_cuda_time": 2.885813590146608, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.888, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047494154339543224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047494154339543224, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 626.611, - "cuda_time_us": 58.142, - "pct_cuda_time": 0.8543951490129091, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.151, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.3075364053273393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.3075364053273393, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.496, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.053137024162063215, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 266.271, - "cuda_time_us": 16.062, - "pct_cuda_time": 0.23603066429509384, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.335, - "pct_cuda_time": 0.034312763113500434, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.383, - "pct_cuda_time": 0.18196785680277341, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 70.888, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2576910552284128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.392, - "pct_cuda_time": 0.22618503205267612, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.471, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.978, - "cuda_time_us": 131.839, - "pct_cuda_time": 1.937370610758366, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 64.458, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.177949075451047, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.16, - "pct_cuda_time": 1.177949075451047, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.853, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.12649433185482306, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.608, - "pct_cuda_time": 0.12649433185482306, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.362, - "cuda_time_us": 43.071, - "pct_cuda_time": 0.6329272034524958, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.071, - "pct_cuda_time": 0.6329272034524958, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 906.361, - "cuda_time_us": 197.086, - "pct_cuda_time": 2.8961735464613914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.34, - "cuda_time_us": 3.263, - "pct_cuda_time": 0.04794969851792374, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.263, - "pct_cuda_time": 0.04794969851792374, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 641.319, - "cuda_time_us": 57.727000000000004, - "pct_cuda_time": 0.8482967350120086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.63, - "cuda_time_us": 20.544, - "pct_cuda_time": 0.3018935355048193, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.544, - "pct_cuda_time": 0.3018935355048193, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.146, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 274.003, - "cuda_time_us": 15.999, - "pct_cuda_time": 0.23510488096483662, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.271, - "pct_cuda_time": 0.03337228480974711, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.416, - "pct_cuda_time": 0.18245279092814626, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 69.638, - "cuda_time_us": 17.535999999999998, - "pct_cuda_time": 0.2576910552284127, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.2, - "pct_cuda_time": 0.22336359714141613, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03432745808699659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 35.938, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.901, - "cuda_time_us": 132.96, - "pct_cuda_time": 1.9538436760475457, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.961, - "cuda_time_us": 80.256, - "pct_cuda_time": 1.1793597929066773, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.256, - "pct_cuda_time": 1.1793597929066773, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.844, - "cuda_time_us": 9.664, - "pct_cuda_time": 0.142012223866753, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.664, - "pct_cuda_time": 0.142012223866753, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.085, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.6324716592741152, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.6324716592741152, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 918.104, - "cuda_time_us": 197.218, - "pct_cuda_time": 2.898113282962882, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.736, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 648.183, - "cuda_time_us": 58.528999999999996, - "pct_cuda_time": 0.8600821037559174, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.654, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.30565544871983263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.30565544871983263, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 192.988, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.05172630670643321, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.52, - "pct_cuda_time": 0.05172630670643321, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 279.576, - "cuda_time_us": 16.256, - "pct_cuda_time": 0.23888148915334612, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.03667865384637991, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.416, - "pct_cuda_time": 0.18245279092814626, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.004, - "cuda_time_us": 17.953, - "pct_cuda_time": 0.26381885917630554, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.808, - "pct_cuda_time": 0.2322981410270728, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.145, - "pct_cuda_time": 0.03152071814923274, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 36.764, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.04609813185740937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.04609813185740937, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.003, - "cuda_time_us": 132.416, - "pct_cuda_time": 1.9458496104656422, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.206, - "cuda_time_us": 80.064, - "pct_cuda_time": 1.176538357995417, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.064, - "pct_cuda_time": 1.176538357995417, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.192, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.1311967233735897, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.1311967233735897, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.44, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.6381145290966351, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.6381145290966351, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 865.657, - "cuda_time_us": 197.05600000000004, - "pct_cuda_time": 2.8957326972565074, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.444, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047494154339543224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047494154339543224, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 605.029, - "cuda_time_us": 58.304, - "pct_cuda_time": 0.8567757347192846, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.114, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.309417361934846, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.309417361934846, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 179.301, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 261.633, - "cuda_time_us": 16.128, - "pct_cuda_time": 0.23700053254583942, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.18386350838377624, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.272, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.2567505769246594, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.2252445537489228, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.508, - "cuda_time_us": 3.199, - "pct_cuda_time": 0.04700922021417041, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.199, - "pct_cuda_time": 0.04700922021417041, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 177.372, - "cuda_time_us": 132.32100000000003, - "pct_cuda_time": 1.9444535879835085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 57.603, - "cuda_time_us": 80.385, - "pct_cuda_time": 1.1812554444876802, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.385, - "pct_cuda_time": 1.1812554444876802, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 46.649, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12978600591795972, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12978600591795972, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.583, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.6334121375778685, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.6334121375778685, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 841.089, - "cuda_time_us": 198.65800000000002, - "pct_cuda_time": 2.919274044797332, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.607, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047964393491419885, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047964393491419885, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 593.741, - "cuda_time_us": 58.272000000000006, - "pct_cuda_time": 0.856305495567408, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.718, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.3075364053273393, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.928, - "pct_cuda_time": 0.3075364053273393, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 176.931, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.054547741617693206, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.054547741617693206, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 255.978, - "cuda_time_us": 16.096, - "pct_cuda_time": 0.2365302933939628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.48, - "pct_cuda_time": 0.18339326923189958, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.179, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.2576910552284128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.296, - "pct_cuda_time": 0.22477431459704614, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03291674063136659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.718, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048434632643296546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048434632643296546, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 166.395, - "cuda_time_us": 133.826, - "pct_cuda_time": 1.9665695230952078, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 57.458, - "cuda_time_us": 80.8, - "pct_cuda_time": 1.1873538584885805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.8, - "pct_cuda_time": 1.1873538584885805, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 35.952, - "cuda_time_us": 8.993, - "pct_cuda_time": 0.13215189665083918, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.993, - "pct_cuda_time": 0.13215189665083918, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.712, - "cuda_time_us": 44.033, - "pct_cuda_time": 0.647063767955788, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.033, - "pct_cuda_time": 0.647063767955788, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 882.41, - "cuda_time_us": 198.84699999999998, - "pct_cuda_time": 2.9220513947881033, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.488, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.046083436883913226, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 626.328, - "cuda_time_us": 58.879, - "pct_cuda_time": 0.8652253444795686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.236, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.3174114275167493, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.6, - "pct_cuda_time": 0.3174114275167493, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.603, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.055488219921446535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.055488219921446535, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 265.918, - "cuda_time_us": 15.999, - "pct_cuda_time": 0.23510488096483662, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.271, - "pct_cuda_time": 0.03337228480974711, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.416, - "pct_cuda_time": 0.18245279092814626, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.865, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.2572208160765361, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.2257147929007995, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.034, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.605, - "cuda_time_us": 133.664, - "pct_cuda_time": 1.964188937388832, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.317, - "cuda_time_us": 81.824, - "pct_cuda_time": 1.2024015113486337, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.824, - "pct_cuda_time": 1.2024015113486337, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.914, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.727, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.6334121375778685, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.6334121375778685, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 955.724, - "cuda_time_us": 196.034, - "pct_cuda_time": 2.8807144343434454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.554, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04797908846491604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.265, - "pct_cuda_time": 0.04797908846491604, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 700.138, - "cuda_time_us": 57.794, - "pct_cuda_time": 0.8492812982362503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.807, - "cuda_time_us": 20.673, - "pct_cuda_time": 0.3037891870858221, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.673, - "pct_cuda_time": 0.3037891870858221, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.295, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 340.948, - "cuda_time_us": 16.097, - "pct_cuda_time": 0.23654498836745896, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.273, - "pct_cuda_time": 0.0334016747567394, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.48, - "pct_cuda_time": 0.18339326923189958, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 71.733, - "cuda_time_us": 17.375999999999998, - "pct_cuda_time": 0.2553398594690294, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.232, - "pct_cuda_time": 0.2238338362932928, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.401, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 174.074, - "cuda_time_us": 131.775, - "pct_cuda_time": 1.9364301324546127, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.672, - "cuda_time_us": 79.904, - "pct_cuda_time": 1.1741871622360338, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.904, - "pct_cuda_time": 1.1741871622360338, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.517, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.1283752884623297, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.515, - "cuda_time_us": 43.135, - "pct_cuda_time": 0.6338676817562491, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.135, - "pct_cuda_time": 0.6338676817562491, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 861.258, - "cuda_time_us": 197.728, - "pct_cuda_time": 2.905607719445917, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.688, - "cuda_time_us": 3.201, - "pct_cuda_time": 0.0470386101611627, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.201, - "pct_cuda_time": 0.0470386101611627, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 610.492, - "cuda_time_us": 58.334999999999994, - "pct_cuda_time": 0.8572312788976651, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.884, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.3042447312642026, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.3042447312642026, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.511, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 258.403, - "cuda_time_us": 16.287, - "pct_cuda_time": 0.23933703333172662, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.03338697978324325, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.703, - "pct_cuda_time": 0.18667024832154006, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.221, - "cuda_time_us": 17.695999999999998, - "pct_cuda_time": 0.26004225098779604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.2280659886601828, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03197626232761326, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.749, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048434632643296546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048434632643296546, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.496, - "cuda_time_us": 132.89600000000002, - "pct_cuda_time": 1.9529031977437925, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.316, - "cuda_time_us": 80.896, - "pct_cuda_time": 1.1887645759442105, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.896, - "pct_cuda_time": 1.1887645759442105, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.767, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.13072648422171304, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.13072648422171304, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.504, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.6334121375778685, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.104, - "pct_cuda_time": 0.6334121375778685, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 932.212, - "cuda_time_us": 197.40699999999998, - "pct_cuda_time": 2.9008906329536535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.46, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.04702391518766656, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 679.495, - "cuda_time_us": 58.462999999999994, - "pct_cuda_time": 0.8591122355051717, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.206, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.3117685576942293, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.3117685576942293, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 205.677, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05360726331393987, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 280.05, - "cuda_time_us": 16.159, - "pct_cuda_time": 0.23745607672421995, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03291674063136659, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.607, - "pct_cuda_time": 0.18525953086591007, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 90.0, - "cuda_time_us": 17.44, - "pct_cuda_time": 0.25628033777278275, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.295, - "pct_cuda_time": 0.22475961962354998, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.145, - "pct_cuda_time": 0.03152071814923274, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.819, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047494154339543224, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.047494154339543224, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.722, - "cuda_time_us": 132.512, - "pct_cuda_time": 1.9472603279212721, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.552, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1835919452735673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.544, - "pct_cuda_time": 1.1835919452735673, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.216, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13213720167734305, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13213720167734305, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.972, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6315311809703618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6315311809703618, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 875.613, - "cuda_time_us": 199.06900000000002, - "pct_cuda_time": 2.9253136789042484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.075, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.0559584590733232, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.808, - "pct_cuda_time": 0.0559584590733232, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 626.574, - "cuda_time_us": 58.687000000000005, - "pct_cuda_time": 0.8624039095683087, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.833, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.3117685576942293, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.216, - "pct_cuda_time": 0.3117685576942293, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 179.109, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.055488219921446535, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.776, - "pct_cuda_time": 0.055488219921446535, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 277.092, - "cuda_time_us": 16.191000000000003, - "pct_cuda_time": 0.23792631587609667, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.303, - "pct_cuda_time": 0.03384252396162377, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.576, - "pct_cuda_time": 0.1848039866875296, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019279805226943288, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.355, - "cuda_time_us": 17.503999999999998, - "pct_cuda_time": 0.2572208160765361, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.22665527120455278, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.03056554487198326, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.43, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.673, - "cuda_time_us": 133.406, - "pct_cuda_time": 1.9603976342268266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.594, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.188294336792334, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.188294336792334, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.083, - "cuda_time_us": 8.991, - "pct_cuda_time": 0.13212250670384687, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.991, - "pct_cuda_time": 0.13212250670384687, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.51, - "cuda_time_us": 43.551, - "pct_cuda_time": 0.6399807907306457, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.551, - "pct_cuda_time": 0.6399807907306457, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 892.194, - "cuda_time_us": 197.344, - "pct_cuda_time": 2.8999648496233963, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.426, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.046553676035789894, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 631.197, - "cuda_time_us": 58.016, - "pct_cuda_time": 0.8525435823523947, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.204, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.3070661661754626, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.896, - "pct_cuda_time": 0.3070661661754626, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 180.861, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.05266678501018654, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 273.575, - "cuda_time_us": 16.288, - "pct_cuda_time": 0.23935172830522278, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03291674063136659, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.704, - "pct_cuda_time": 0.18668494329503624, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.019750044378819956, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 17], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 64.548, - "cuda_time_us": 17.247999999999998, - "pct_cuda_time": 0.25345890286152273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.104, - "pct_cuda_time": 0.22195287968578614, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.03150602317573659, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.336, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048434632643296546, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.048434632643296546, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.967, - "cuda_time_us": 132.86399999999998, - "pct_cuda_time": 1.952432958591915, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.648, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.182181227817937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.448, - "pct_cuda_time": 1.182181227817937, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.921, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12790504931045305, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.704, - "pct_cuda_time": 0.12790504931045305, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.738, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6423466814635251, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.712, - "pct_cuda_time": 0.6423466814635251, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.41, - "cuda_time_us": 3.201, - "pct_cuda_time": 0.0470386101611627, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.201, - "pct_cuda_time": 0.0470386101611627, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 151.323, - "cuda_time_us": 347.265, - "pct_cuda_time": 5.10304997113907, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.041, - "pct_cuda_time": 0.044687414401779375, - "trace": "index_select(bfloat16[2, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010815500493163308, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 343.488, - "pct_cuda_time": 5.047547056244128, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 1142.125, - "cuda_time_us": 121.568, - "pct_cuda_time": 1.7864385379794525, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03244650147948993, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03244650147948993, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.03291674063136659, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.03244650147948993, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03197626232761326, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.863, - "pct_cuda_time": 0.012681762127173826, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.8, - "pct_cuda_time": 0.01175597879691664, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.128, - "pct_cuda_time": 0.06066085059208986, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.8, - "pct_cuda_time": 0.07053587278149982, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.432, - "pct_cuda_time": 0.5059773274192922, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 28.032, - "pct_cuda_time": 0.41192949704395904, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 1.984, - "pct_cuda_time": 0.029154827416353262, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.928, - "pct_cuda_time": 0.0724168293890065, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 28.064, - "pct_cuda_time": 0.41239973619583575, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.497, - "pct_cuda_time": 0.03669334881987606, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - } -} \ No newline at end of file diff --git a/H100_llama8b_pp1_tp1/profiling_bs2_pl512.json b/H100_llama8b_pp1_tp1/profiling_bs2_pl512.json deleted file mode 100644 index ec1ffc67fc3ab60e4948b302cd6f5a4fcc8d3cd0..0000000000000000000000000000000000000000 --- a/H100_llama8b_pp1_tp1/profiling_bs2_pl512.json +++ /dev/null @@ -1,18219 +0,0 @@ -{ - "context": { - "python_version": "3.12.9 | packaged by Anaconda, Inc. | (main, Feb 6 2025, 18:56:27) [GCC 11.2.0]", - "torch_version": "2.5.1+cu124", - "engine_args": { - "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "served_model_name": null, - "tokenizer": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B", - "task": "auto", - "skip_tokenizer_init": false, - "tokenizer_mode": "auto", - "trust_remote_code": false, - "allowed_local_media_path": null, - "download_dir": null, - "load_format": "dummy", - "config_format": "auto", - "dtype": "auto", - "kv_cache_dtype": "auto", - "seed": 0, - "max_model_len": null, - "distributed_executor_backend": null, - "pipeline_parallel_size": 1, - "tensor_parallel_size": 1, - "max_parallel_loading_workers": null, - "block_size": null, - "enable_prefix_caching": false, - "disable_sliding_window": false, - "use_v2_block_manager": true, - "swap_space": 4, - "cpu_offload_gb": 0, - "gpu_memory_utilization": 0.9, - "max_num_batched_tokens": 8000, - "max_num_partial_prefills": 1, - "max_long_partial_prefills": 1, - "long_prefill_token_threshold": 0, - "max_num_seqs": 256, - "max_logprobs": 20, - "disable_log_stats": false, - "revision": null, - "code_revision": null, - "rope_scaling": null, - "rope_theta": null, - "hf_overrides": null, - "tokenizer_revision": null, - "quantization": null, - "enforce_eager": true, - "max_seq_len_to_capture": 8192, - "disable_custom_all_reduce": false, - "tokenizer_pool_size": 0, - "tokenizer_pool_type": "ray", - "tokenizer_pool_extra_config": null, - "limit_mm_per_prompt": null, - "mm_processor_kwargs": null, - "disable_mm_preprocessor_cache": false, - "enable_lora": false, - "enable_lora_bias": false, - "max_loras": 1, - "max_lora_rank": 16, - "enable_prompt_adapter": false, - "max_prompt_adapters": 1, - "max_prompt_adapter_token": 0, - "fully_sharded_loras": false, - "lora_extra_vocab_size": 256, - "long_lora_scaling_factors": null, - "lora_dtype": "auto", - "max_cpu_loras": null, - "device": "auto", - "num_scheduler_steps": 1, - "multi_step_stream_outputs": true, - "ray_workers_use_nsight": false, - "num_gpu_blocks_override": null, - "num_lookahead_slots": 0, - "model_loader_extra_config": null, - "ignore_patterns": [], - "preemption_mode": null, - "scheduler_delay_factor": 0.0, - "enable_chunked_prefill": null, - "guided_decoding_backend": "xgrammar", - "logits_processor_pattern": null, - "speculative_model": null, - "speculative_model_quantization": null, - "speculative_draft_tensor_parallel_size": null, - "num_speculative_tokens": null, - "speculative_disable_mqa_scorer": false, - "speculative_max_model_len": null, - "speculative_disable_by_batch_size": null, - "ngram_prompt_lookup_max": null, - "ngram_prompt_lookup_min": null, - "spec_decoding_acceptance_method": "rejection_sampler", - "typical_acceptance_sampler_posterior_threshold": null, - "typical_acceptance_sampler_posterior_alpha": null, - "qlora_adapter_name_or_path": null, - "disable_logprobs_during_spec_decoding": null, - "otlp_traces_endpoint": null, - "collect_detailed_traces": null, - "disable_async_output_proc": false, - "scheduling_policy": "fcfs", - "scheduler_cls": "vllm.core.scheduler.Scheduler", - "override_neuron_config": null, - "override_pooler_config": null, - "compilation_config": null, - "worker_cls": "auto", - "kv_transfer_config": null, - "generation_config": null, - "override_generation_config": null, - "enable_sleep_mode": false, - "model_impl": "auto", - "calculate_kv_scales": false, - "additional_config": null - }, - "prompt_len": 0, - "batch_size": 2, - "num_steps": 2, - "complete_num_requests_per_step": null, - "save_chrome_traces_folder": null - }, - "prefill": { - "metadata": { - "num_running_seqs": null - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 22661.483999999997, - "pct_cuda_time": 97.92498616241166, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 29.952, - "pct_cuda_time": 0.12942882229321587, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cuda_time_us": 29.952, - "pct_cuda_time": 0.12942882229321587, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 22620.842999999997, - "pct_cuda_time": 97.74936794770751, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 669.6600000000001, - "pct_cuda_time": 2.8937401554779294, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 13.152, - "pct_cuda_time": 0.056832527737726186, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 656.508, - "pct_cuda_time": 2.836907627740203, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 5576.917999999998, - "pct_cuda_time": 24.099022728560247, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 2674.3289999999997, - "pct_cuda_time": 11.556331894183815, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 24.641000000000002, - "pct_cuda_time": 0.10647888655606076, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2649.6880000000006, - "pct_cuda_time": 11.449853007627757, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 424.927, - "pct_cuda_time": 1.8361979557488428, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 424.927, - "pct_cuda_time": 1.8361979557488428, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 953.4719999999999, - "pct_cuda_time": 4.1201508430007046, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 179.74700000000007, - "pct_cuda_time": 0.7767241760396193, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cuda_time_us": 728.4449999999999, - "pct_cuda_time": 3.147762368302004, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 45.28, - "pct_cuda_time": 0.19566429865908166, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 1524.19, - "pct_cuda_time": 6.586342035626893, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1524.19, - "pct_cuda_time": 6.586342035626893, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 16374.264999999996, - "pct_cuda_time": 70.75660506366933, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 10167.464999999998, - "pct_cuda_time": 43.93573119182331, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 23.67700000000001, - "pct_cuda_time": 0.10231324203513864, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 10143.788, - "pct_cuda_time": 43.83341794978818, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 1445.7199999999998, - "pct_cuda_time": 6.247256843140625, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 1445.7199999999998, - "pct_cuda_time": 6.247256843140625, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 4761.080000000002, - "pct_cuda_time": 20.573617028705407, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 4761.080000000002, - "pct_cuda_time": 20.573617028705407, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 10.689, - "pct_cuda_time": 0.0461893924109303, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 10.689, - "pct_cuda_time": 0.0461893924109303, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 356.44800000000004, - "pct_cuda_time": 1.5402859524830466, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.233, - "pct_cuda_time": 0.013970465493922505, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0031760878868026723, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 352.48, - "pct_cuda_time": 1.5231393991023212, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 123.745, - "pct_cuda_time": 0.5347278851053017, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 16.512, - "pct_cuda_time": 0.07135178664882413, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.096, - "pct_cuda_time": 0.017699668005909858, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.608, - "pct_cuda_time": 0.01991212650664859, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 33.696, - "pct_cuda_time": 0.14560742507986782, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.392, - "pct_cuda_time": 0.1183665297895222, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 1.889, - "pct_cuda_time": 0.008162761929483332, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.672, - "pct_cuda_time": 0.020188683819240934, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 27.84, - "pct_cuda_time": 0.12030243097766859, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 3.04, - "pct_cuda_time": 0.013136472348136224, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 30893.576, - "cuda_time_us": 22661.483999999997, - "pct_cuda_time": 97.92498616241166, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 108.651, - "cuda_time_us": 29.952, - "pct_cuda_time": 0.12942882229321587, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectLargeIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 29.952, - "pct_cuda_time": 0.12942882229321587, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[1024]) <- embedding(bfloat16[128256, 4096], int64[1024], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1495.032, - "cuda_time_us": 711.614, - "pct_cuda_time": 3.0750321162982273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 87.162, - "cuda_time_us": 13.152, - "pct_cuda_time": 0.056832527737726186, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.152, - "pct_cuda_time": 0.056832527737726186, - "trace": "_C::rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 1092.396, - "cuda_time_us": 176.704, - "pct_cuda_time": 0.7635747400674551, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 138.62, - "cuda_time_us": 86.336, - "pct_cuda_time": 0.3730758146870688, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0033186877511080987, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 85.568, - "pct_cuda_time": 0.36975712693596063, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 354.716, - "cuda_time_us": 12.896, - "pct_cuda_time": 0.05572629848735683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.896, - "pct_cuda_time": 0.05572629848735683, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 390.603, - "cuda_time_us": 30.112000000000002, - "pct_cuda_time": 0.13012021557469672, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.664, - "pct_cuda_time": 0.02447532216442223, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.784, - "pct_cuda_time": 0.0984544032828736, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.0071904901274008805, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 108.303, - "cuda_time_us": 47.36, - "pct_cuda_time": 0.20465241131833273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.36, - "pct_cuda_time": 0.20465241131833273, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 43.421, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 217.338, - "cuda_time_us": 511.26200000000006, - "pct_cuda_time": 2.2092694492279024, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 76.535, - "cuda_time_us": 317.343, - "pct_cuda_time": 1.3713051132811165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0031760878868026723, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.608, - "pct_cuda_time": 1.3681290253943137, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 51.66, - "cuda_time_us": 45.216, - "pct_cuda_time": 0.1953877413464893, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.216, - "pct_cuda_time": 0.1953877413464893, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 61.491, - "cuda_time_us": 148.703, - "pct_cuda_time": 0.6425765946002964, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.703, - "pct_cuda_time": 0.6425765946002964, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 992.739, - "cuda_time_us": 706.302, - "pct_cuda_time": 3.052077859353063, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.081, - "cuda_time_us": 10.464, - "pct_cuda_time": 0.04521712060884785, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.464, - "pct_cuda_time": 0.04521712060884785, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 715.418, - "cuda_time_us": 174.336, - "pct_cuda_time": 0.7533421195015384, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 64.159, - "cuda_time_us": 83.13600000000001, - "pct_cuda_time": 0.3592479490574517, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.4, - "pct_cuda_time": 0.3560675399626398, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 208.977, - "cuda_time_us": 13.408, - "pct_cuda_time": 0.05793875698809556, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.408, - "pct_cuda_time": 0.05793875698809556, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 311.389, - "cuda_time_us": 30.016000000000002, - "pct_cuda_time": 0.1297053796058082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.888, - "pct_cuda_time": 0.025443272758495422, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.816, - "pct_cuda_time": 0.09859268193916976, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.005669424908143003, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.076, - "cuda_time_us": 47.776, - "pct_cuda_time": 0.20645003385018298, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.776, - "pct_cuda_time": 0.20645003385018298, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.333, - "cuda_time_us": 10.591, - "pct_cuda_time": 0.04576591402602327, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.591, - "pct_cuda_time": 0.04576591402602327, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.241, - "cuda_time_us": 510.911, - "pct_cuda_time": 2.2077527052166532, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.276, - "cuda_time_us": 317.471, - "pct_cuda_time": 1.371858227906301, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.735, - "pct_cuda_time": 1.3686778188114892, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 40.321, - "cuda_time_us": 45.344, - "pct_cuda_time": 0.19594085597167402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.344, - "pct_cuda_time": 0.19594085597167402, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.311, - "cuda_time_us": 148.096, - "pct_cuda_time": 0.6399536213386784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.096, - "pct_cuda_time": 0.6399536213386784, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 973.163, - "cuda_time_us": 705.951, - "pct_cuda_time": 3.0505611153418144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.173, - "cuda_time_us": 10.592, - "pct_cuda_time": 0.04577023523403253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.592, - "pct_cuda_time": 0.04577023523403253, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 688.762, - "cuda_time_us": 174.01600000000002, - "pct_cuda_time": 0.7519593329385768, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 60.498, - "cuda_time_us": 83.424, - "pct_cuda_time": 0.36049245696411725, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.688, - "pct_cuda_time": 0.3573120478693053, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 194.504, - "cuda_time_us": 13.728, - "pct_cuda_time": 0.05932154355105726, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.728, - "pct_cuda_time": 0.05932154355105726, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 299.485, - "cuda_time_us": 29.663999999999998, - "pct_cuda_time": 0.1281843143865503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.568, - "pct_cuda_time": 0.024060486195533713, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.753, - "pct_cuda_time": 0.09832044583458668, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.005803382356429917, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 76.298, - "cuda_time_us": 47.2, - "pct_cuda_time": 0.2039610180368519, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.2, - "pct_cuda_time": 0.2039610180368519, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 34.756, - "cuda_time_us": 10.336, - "pct_cuda_time": 0.044664005983663164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.336, - "pct_cuda_time": 0.044664005983663164, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 185.137, - "cuda_time_us": 511.00699999999995, - "pct_cuda_time": 2.2081675411855417, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.269, - "cuda_time_us": 317.823, - "pct_cuda_time": 1.3733792931255588, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.087, - "pct_cuda_time": 1.3701988840307469, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 40.818, - "cuda_time_us": 44.512, - "pct_cuda_time": 0.19234561090797356, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.512, - "pct_cuda_time": 0.19234561090797356, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.387, - "cuda_time_us": 148.672, - "pct_cuda_time": 0.6424426371520094, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.672, - "pct_cuda_time": 0.6424426371520094, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 950.793, - "cuda_time_us": 704.83, - "pct_cuda_time": 3.045717041163439, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.132, - "cuda_time_us": 10.08, - "pct_cuda_time": 0.04355777673329379, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.08, - "pct_cuda_time": 0.04355777673329379, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 683.677, - "cuda_time_us": 173.75900000000001, - "pct_cuda_time": 0.7508487824801982, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 61.858, - "cuda_time_us": 83.423, - "pct_cuda_time": 0.36048813575610794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.687, - "pct_cuda_time": 0.35730772666129607, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 196.978, - "cuda_time_us": 13.152, - "pct_cuda_time": 0.056832527737726186, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.152, - "pct_cuda_time": 0.056832527737726186, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 295.976, - "cuda_time_us": 29.824, - "pct_cuda_time": 0.1288757076680312, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.6, - "pct_cuda_time": 0.024198764851829885, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.72, - "pct_cuda_time": 0.09817784597028124, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 70.792, - "cuda_time_us": 47.36, - "pct_cuda_time": 0.20465241131833273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.36, - "pct_cuda_time": 0.20465241131833273, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.739, - "cuda_time_us": 10.336, - "pct_cuda_time": 0.044664005983663164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.336, - "pct_cuda_time": 0.044664005983663164, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.687, - "cuda_time_us": 510.655, - "pct_cuda_time": 2.2066464759662843, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.657, - "cuda_time_us": 316.703, - "pct_cuda_time": 1.3685395401551927, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 315.967, - "pct_cuda_time": 1.365359131060381, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.639, - "cuda_time_us": 45.376, - "pct_cuda_time": 0.19607913462797014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.376, - "pct_cuda_time": 0.19607913462797014, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.074, - "cuda_time_us": 148.576, - "pct_cuda_time": 0.642027801183121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.576, - "pct_cuda_time": 0.642027801183121, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 933.965, - "cuda_time_us": 706.9100000000001, - "pct_cuda_time": 3.054705153822691, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.934, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 665.499, - "cuda_time_us": 174.43200000000002, - "pct_cuda_time": 0.7537569554704271, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.968, - "cuda_time_us": 83.77600000000001, - "pct_cuda_time": 0.3620135221833751, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 83.04, - "pct_cuda_time": 0.3588331130885632, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.137, - "cuda_time_us": 13.728, - "pct_cuda_time": 0.05932154355105726, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.728, - "pct_cuda_time": 0.05932154355105726, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 294.426, - "cuda_time_us": 29.856, - "pct_cuda_time": 0.12901398632432737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.601, - "pct_cuda_time": 0.024203086059839143, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.783, - "pct_cuda_time": 0.09845008207486435, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.0063608181896238555, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 73.249, - "cuda_time_us": 47.072, - "pct_cuda_time": 0.20340790341166726, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.072, - "pct_cuda_time": 0.20340790341166726, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.817, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.232, - "cuda_time_us": 511.486, - "pct_cuda_time": 2.2102373998219753, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.12, - "cuda_time_us": 317.695, - "pct_cuda_time": 1.3728261785003744, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.959, - "pct_cuda_time": 1.3696457694055624, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.848, - "cuda_time_us": 44.799, - "pct_cuda_time": 0.19358579760662983, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.799, - "pct_cuda_time": 0.19358579760662983, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.777, - "cuda_time_us": 148.992, - "pct_cuda_time": 0.6438254237149711, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.992, - "pct_cuda_time": 0.6438254237149711, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 983.623, - "cuda_time_us": 706.686, - "pct_cuda_time": 3.053737203228617, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.842, - "cuda_time_us": 10.592, - "pct_cuda_time": 0.04577023523403253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.592, - "pct_cuda_time": 0.04577023523403253, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 705.229, - "cuda_time_us": 174.04700000000003, - "pct_cuda_time": 0.7520932903868637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 59.483, - "cuda_time_us": 83.328, - "pct_cuda_time": 0.3600776209952287, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.592, - "pct_cuda_time": 0.3568972119004168, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 221.057, - "cuda_time_us": 12.992, - "pct_cuda_time": 0.056141134456245335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.992, - "pct_cuda_time": 0.056141134456245335, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 291.958, - "cuda_time_us": 29.471, - "pct_cuda_time": 0.12735032124076404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.568, - "pct_cuda_time": 0.024060486195533713, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.591, - "pct_cuda_time": 0.09762041013708732, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.005669424908143003, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 69.83, - "cuda_time_us": 48.256, - "pct_cuda_time": 0.20852421369462554, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 48.256, - "pct_cuda_time": 0.20852421369462554, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.232, - "cuda_time_us": 10.784, - "pct_cuda_time": 0.046599907171809556, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.784, - "pct_cuda_time": 0.046599907171809556, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.65, - "cuda_time_us": 511.26300000000003, - "pct_cuda_time": 2.2092737704359116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.592, - "cuda_time_us": 317.664, - "pct_cuda_time": 1.3726922210520873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.928, - "pct_cuda_time": 1.3695118119572753, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.287, - "cuda_time_us": 45.119, - "pct_cuda_time": 0.19496858416959154, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.119, - "pct_cuda_time": 0.19496858416959154, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.079, - "cuda_time_us": 148.48, - "pct_cuda_time": 0.6416129652142324, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.48, - "pct_cuda_time": 0.6416129652142324, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 942.802, - "cuda_time_us": 706.206, - "pct_cuda_time": 3.0516630233841746, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.279, - "cuda_time_us": 10.432, - "pct_cuda_time": 0.04507884195255167, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.432, - "pct_cuda_time": 0.04507884195255167, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 670.878, - "cuda_time_us": 173.85500000000002, - "pct_cuda_time": 0.7512636184490866, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.032, - "cuda_time_us": 83.39200000000001, - "pct_cuda_time": 0.36035417830782107, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.656, - "pct_cuda_time": 0.3571737692130092, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 191.669, - "cuda_time_us": 13.056, - "pct_cuda_time": 0.05641769176883768, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.056, - "pct_cuda_time": 0.05641769176883768, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 288.981, - "cuda_time_us": 29.983000000000004, - "pct_cuda_time": 0.1295627797415028, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.824, - "pct_cuda_time": 0.02516671544590308, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.815, - "pct_cuda_time": 0.09858836073116052, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.005807703564439173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 74.51, - "cuda_time_us": 47.424, - "pct_cuda_time": 0.20492896863092508, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.424, - "pct_cuda_time": 0.20492896863092508, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.558, - "cuda_time_us": 10.464, - "pct_cuda_time": 0.04521712060884785, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.464, - "pct_cuda_time": 0.04521712060884785, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.421, - "cuda_time_us": 511.455, - "pct_cuda_time": 2.210103442373688, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.756, - "cuda_time_us": 317.664, - "pct_cuda_time": 1.3726922210520873, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.928, - "pct_cuda_time": 1.3695118119572753, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.07, - "cuda_time_us": 45.279, - "pct_cuda_time": 0.19565997745107241, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.279, - "pct_cuda_time": 0.19565997745107241, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.291, - "cuda_time_us": 148.512, - "pct_cuda_time": 0.6417512438705286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.512, - "pct_cuda_time": 0.6417512438705286, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 929.938, - "cuda_time_us": 706.3979999999999, - "pct_cuda_time": 3.052492695321951, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.553, - "cuda_time_us": 10.048, - "pct_cuda_time": 0.043419498076997624, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.048, - "pct_cuda_time": 0.043419498076997624, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 662.074, - "cuda_time_us": 173.952, - "pct_cuda_time": 0.7516827756259843, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 59.381, - "cuda_time_us": 83.52000000000001, - "pct_cuda_time": 0.3609072929330058, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.784, - "pct_cuda_time": 0.35772688383819384, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 196.746, - "cuda_time_us": 12.896, - "pct_cuda_time": 0.05572629848735683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.896, - "pct_cuda_time": 0.05572629848735683, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 275.017, - "cuda_time_us": 30.080000000000002, - "pct_cuda_time": 0.12998193691840054, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.632, - "pct_cuda_time": 0.024337043508126058, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.944, - "pct_cuda_time": 0.09914579656435443, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 75.141, - "cuda_time_us": 47.456, - "pct_cuda_time": 0.20506724728722125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.456, - "pct_cuda_time": 0.20506724728722125, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.551, - "cuda_time_us": 10.079, - "pct_cuda_time": 0.04355345552528454, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.079, - "pct_cuda_time": 0.04355345552528454, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 180.57, - "cuda_time_us": 512.319, - "pct_cuda_time": 2.2138369660936847, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.286, - "cuda_time_us": 317.53599999999994, - "pct_cuda_time": 1.3721391064269024, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0033186877511080987, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.768, - "pct_cuda_time": 1.3688204186757944, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.376, - "cuda_time_us": 45.311, - "pct_cuda_time": 0.1957982561073686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.311, - "pct_cuda_time": 0.1957982561073686, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.999, - "cuda_time_us": 149.472, - "pct_cuda_time": 0.6458996035594138, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.472, - "pct_cuda_time": 0.6458996035594138, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 906.355, - "cuda_time_us": 707.486, - "pct_cuda_time": 3.057194169636021, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.12, - "cuda_time_us": 10.144, - "pct_cuda_time": 0.04383433404588614, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.144, - "pct_cuda_time": 0.04383433404588614, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 645.142, - "cuda_time_us": 174.144, - "pct_cuda_time": 0.7525124475637613, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.601, - "cuda_time_us": 82.912, - "pct_cuda_time": 0.35827999846337855, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.176, - "pct_cuda_time": 0.35509958936856656, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 182.652, - "cuda_time_us": 13.376, - "pct_cuda_time": 0.05780047833179938, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.376, - "pct_cuda_time": 0.05780047833179938, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 273.888, - "cuda_time_us": 29.76, - "pct_cuda_time": 0.12859915035543884, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.632, - "pct_cuda_time": 0.024337043508126058, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.09776301000139274, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.61, - "cuda_time_us": 48.096, - "pct_cuda_time": 0.20783282041314466, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 48.096, - "pct_cuda_time": 0.20783282041314466, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.362, - "cuda_time_us": 10.688, - "pct_cuda_time": 0.04618507120292104, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.688, - "pct_cuda_time": 0.04618507120292104, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.74, - "cuda_time_us": 512.51, - "pct_cuda_time": 2.214662316823453, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.427, - "cuda_time_us": 317.599, - "pct_cuda_time": 1.3724113425314857, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0031760878868026723, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.864, - "pct_cuda_time": 1.369235254644683, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.725, - "cuda_time_us": 45.376, - "pct_cuda_time": 0.19607913462797014, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.376, - "pct_cuda_time": 0.19607913462797014, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 57.92, - "cuda_time_us": 149.535, - "pct_cuda_time": 0.6461718396639968, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.535, - "pct_cuda_time": 0.6461718396639968, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 929.035, - "cuda_time_us": 706.3040000000001, - "pct_cuda_time": 3.052086501769082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.451, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 650.164, - "cuda_time_us": 174.497, - "pct_cuda_time": 0.7540378339910286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.375, - "cuda_time_us": 83.48899999999999, - "pct_cuda_time": 0.3607733354847188, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0031847303028211833, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.752, - "pct_cuda_time": 0.3575886051818976, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 182.104, - "cuda_time_us": 13.12, - "pct_cuda_time": 0.05669424908143002, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.12, - "pct_cuda_time": 0.05669424908143002, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 283.912, - "cuda_time_us": 29.6, - "pct_cuda_time": 0.12790775707395796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.536, - "pct_cuda_time": 0.023922207539237544, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.784, - "pct_cuda_time": 0.0984544032828736, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.005531146251846831, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.021, - "cuda_time_us": 48.288, - "pct_cuda_time": 0.2086624923509217, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 48.288, - "pct_cuda_time": 0.2086624923509217, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.678, - "cuda_time_us": 10.784, - "pct_cuda_time": 0.046599907171809556, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.784, - "pct_cuda_time": 0.046599907171809556, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 188.509, - "cuda_time_us": 510.52700000000004, - "pct_cuda_time": 2.2060933613411, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 71.209, - "cuda_time_us": 317.759, - "pct_cuda_time": 1.3731027358129666, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.023, - "pct_cuda_time": 1.3699223267181548, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.839, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.19414323343982376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.19414323343982376, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.843, - "cuda_time_us": 147.84, - "pct_cuda_time": 0.638847392088309, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 147.84, - "pct_cuda_time": 0.638847392088309, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 902.944, - "cuda_time_us": 705.31, - "pct_cuda_time": 3.0477912210078815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.432, - "cuda_time_us": 10.368, - "pct_cuda_time": 0.04480228463995933, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.368, - "pct_cuda_time": 0.04480228463995933, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 639.143, - "cuda_time_us": 173.887, - "pct_cuda_time": 0.7514018971053827, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.253, - "cuda_time_us": 83.135, - "pct_cuda_time": 0.35924362784944247, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0031760878868026723, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.4, - "pct_cuda_time": 0.3560675399626398, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.068, - "cuda_time_us": 13.375, - "pct_cuda_time": 0.057796157123790125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.375, - "pct_cuda_time": 0.057796157123790125, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 274.509, - "cuda_time_us": 29.729, - "pct_cuda_time": 0.12846519290715191, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.569, - "pct_cuda_time": 0.02406480740354297, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.816, - "pct_cuda_time": 0.09859268193916976, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.005807703564439173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.353, - "cuda_time_us": 47.648, - "pct_cuda_time": 0.2058969192249983, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.648, - "pct_cuda_time": 0.2058969192249983, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.892, - "cuda_time_us": 10.24, - "pct_cuda_time": 0.04424917001477465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.24, - "pct_cuda_time": 0.04424917001477465, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 177.995, - "cuda_time_us": 510.81499999999994, - "pct_cuda_time": 2.2073378692477648, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.011, - "cuda_time_us": 317.663, - "pct_cuda_time": 1.3726878998440781, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.927, - "pct_cuda_time": 1.3695074907492661, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 45.751, - "cuda_time_us": 45.184, - "pct_cuda_time": 0.19524946269019314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.184, - "pct_cuda_time": 0.19524946269019314, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.261, - "cuda_time_us": 147.968, - "pct_cuda_time": 0.6394005067134937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 147.968, - "pct_cuda_time": 0.6394005067134937, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 972.828, - "cuda_time_us": 705.9179999999999, - "pct_cuda_time": 3.0504185154775083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.173, - "cuda_time_us": 10.112, - "pct_cuda_time": 0.04369605538958996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.112, - "pct_cuda_time": 0.04369605538958996, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 696.718, - "cuda_time_us": 174.94199999999998, - "pct_cuda_time": 0.7559607715551471, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 61.11, - "cuda_time_us": 84.70299999999999, - "pct_cuda_time": 0.36601928200795475, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 1.728, - "pct_cuda_time": 0.0074670474399932225, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.975, - "pct_cuda_time": 0.35855223456796154, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 198.074, - "cuda_time_us": 13.024, - "pct_cuda_time": 0.056279413112541504, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.024, - "pct_cuda_time": 0.056279413112541504, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 289.715, - "cuda_time_us": 29.791, - "pct_cuda_time": 0.12873310780372574, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.503, - "pct_cuda_time": 0.02377960767493212, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.784, - "pct_cuda_time": 0.0984544032828736, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.524, - "cuda_time_us": 47.424, - "pct_cuda_time": 0.20492896863092508, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.424, - "pct_cuda_time": 0.20492896863092508, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.245, - "cuda_time_us": 10.272, - "pct_cuda_time": 0.04438744867107082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.272, - "pct_cuda_time": 0.04438744867107082, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 182.937, - "cuda_time_us": 510.592, - "pct_cuda_time": 2.206374239861701, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.925, - "cuda_time_us": 317.408, - "pct_cuda_time": 1.371585991801718, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.672, - "pct_cuda_time": 1.3684055827069062, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 46.836, - "cuda_time_us": 44.8, - "pct_cuda_time": 0.19359011881463908, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.8, - "pct_cuda_time": 0.19359011881463908, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.166, - "cuda_time_us": 148.384, - "pct_cuda_time": 0.6411981292453439, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.384, - "pct_cuda_time": 0.6411981292453439, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 930.534, - "cuda_time_us": 707.932, - "pct_cuda_time": 3.059121428408149, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.094, - "cuda_time_us": 10.559, - "pct_cuda_time": 0.0456276353697271, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.559, - "pct_cuda_time": 0.0456276353697271, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 666.605, - "cuda_time_us": 174.976, - "pct_cuda_time": 0.7561076926274618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.902, - "cuda_time_us": 83.39200000000001, - "pct_cuda_time": 0.36035417830782107, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.656, - "pct_cuda_time": 0.3571737692130092, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 187.158, - "cuda_time_us": 13.856, - "pct_cuda_time": 0.05987465817624194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.856, - "pct_cuda_time": 0.05987465817624194, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 301.116, - "cuda_time_us": 29.887999999999998, - "pct_cuda_time": 0.12915226498062352, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.696, - "pct_cuda_time": 0.0246136008207184, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.72, - "pct_cuda_time": 0.09817784597028124, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.0063608181896238555, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.422, - "cuda_time_us": 47.84, - "pct_cuda_time": 0.20672659116277534, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.84, - "pct_cuda_time": 0.20672659116277534, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.208, - "cuda_time_us": 10.495, - "pct_cuda_time": 0.04535107805713476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.495, - "pct_cuda_time": 0.04535107805713476, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.995, - "cuda_time_us": 511.902, - "pct_cuda_time": 2.2120350223538257, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.772, - "cuda_time_us": 317.85499999999996, - "pct_cuda_time": 1.373517571781855, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0033186877511080987, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.087, - "pct_cuda_time": 1.3701988840307469, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.465, - "cuda_time_us": 45.088, - "pct_cuda_time": 0.19483462672130464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.088, - "pct_cuda_time": 0.19483462672130464, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 62.617, - "cuda_time_us": 148.959, - "pct_cuda_time": 0.6436828238506658, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.959, - "pct_cuda_time": 0.6436828238506658, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 878.565, - "cuda_time_us": 706.8149999999999, - "pct_cuda_time": 3.054294639061811, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.152, - "cuda_time_us": 10.431, - "pct_cuda_time": 0.04507452074454242, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.431, - "pct_cuda_time": 0.04507452074454242, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 623.543, - "cuda_time_us": 174.20899999999997, - "pct_cuda_time": 0.7527933260843629, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.489, - "cuda_time_us": 83.74499999999999, - "pct_cuda_time": 0.36187956473508814, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0031847303028211833, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 83.008, - "pct_cuda_time": 0.35869483443226696, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 178.472, - "cuda_time_us": 13.472, - "pct_cuda_time": 0.058215314300687895, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.472, - "pct_cuda_time": 0.058215314300687895, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 272.167, - "cuda_time_us": 29.471999999999998, - "pct_cuda_time": 0.12735464244877326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.536, - "pct_cuda_time": 0.023922207539237544, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.624, - "pct_cuda_time": 0.09776301000139274, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.005669424908143003, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.31, - "cuda_time_us": 47.52, - "pct_cuda_time": 0.2053438045998136, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.52, - "pct_cuda_time": 0.2053438045998136, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.862, - "cuda_time_us": 10.4, - "pct_cuda_time": 0.0449405632962555, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.4, - "pct_cuda_time": 0.0449405632962555, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.667, - "cuda_time_us": 511.775, - "pct_cuda_time": 2.21148622893665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.958, - "cuda_time_us": 317.407, - "pct_cuda_time": 1.3715816705937087, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.671, - "pct_cuda_time": 1.3684012614988967, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.916, - "cuda_time_us": 45.056, - "pct_cuda_time": 0.19469634806500846, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.056, - "pct_cuda_time": 0.19469634806500846, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.03, - "cuda_time_us": 149.312, - "pct_cuda_time": 0.6452082102779328, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.312, - "pct_cuda_time": 0.6452082102779328, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 908.945, - "cuda_time_us": 705.2149999999999, - "pct_cuda_time": 3.047380706247002, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.375, - "cuda_time_us": 10.656, - "pct_cuda_time": 0.04604679254662487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.656, - "pct_cuda_time": 0.04604679254662487, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 659.26, - "cuda_time_us": 173.632, - "pct_cuda_time": 0.7502999890630226, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 63.859, - "cuda_time_us": 83.19999999999999, - "pct_cuda_time": 0.35952450637004396, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0031847303028211833, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.463, - "pct_cuda_time": 0.35633977606722284, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 193.994, - "cuda_time_us": 13.344, - "pct_cuda_time": 0.05766219967550322, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.344, - "pct_cuda_time": 0.05766219967550322, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 281.925, - "cuda_time_us": 29.728, - "pct_cuda_time": 0.12846087169914266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.504, - "pct_cuda_time": 0.023783928882941372, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.09886923925176211, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.005807703564439173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.106, - "cuda_time_us": 47.36, - "pct_cuda_time": 0.20465241131833273, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.36, - "pct_cuda_time": 0.20465241131833273, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.832, - "cuda_time_us": 9.92, - "pct_cuda_time": 0.04286638345181294, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.92, - "pct_cuda_time": 0.04286638345181294, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.516, - "cuda_time_us": 511.00699999999995, - "pct_cuda_time": 2.2081675411855417, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.277, - "cuda_time_us": 317.311, - "pct_cuda_time": 1.3711668346248203, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.575, - "pct_cuda_time": 1.3679864255300083, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.452, - "cuda_time_us": 45.472, - "pct_cuda_time": 0.19649397059685866, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.472, - "pct_cuda_time": 0.19649397059685866, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.154, - "cuda_time_us": 148.224, - "pct_cuda_time": 0.640506735963863, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.224, - "pct_cuda_time": 0.640506735963863, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 897.724, - "cuda_time_us": 705.8509999999999, - "pct_cuda_time": 3.0501289945408883, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.385, - "cuda_time_us": 10.24, - "pct_cuda_time": 0.04424917001477465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.24, - "pct_cuda_time": 0.04424917001477465, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 638.891, - "cuda_time_us": 174.397, - "pct_cuda_time": 0.7536057131901029, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.844, - "cuda_time_us": 83.199, - "pct_cuda_time": 0.35952018516203477, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.463, - "pct_cuda_time": 0.35633977606722284, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 188.495, - "cuda_time_us": 13.151, - "pct_cuda_time": 0.05682820652971694, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.151, - "pct_cuda_time": 0.05682820652971694, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 276.424, - "cuda_time_us": 30.302999999999997, - "pct_cuda_time": 0.13094556630446447, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.792, - "pct_cuda_time": 0.025028436789606912, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 23.008, - "pct_cuda_time": 0.09942235387694678, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.006494775637910771, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 64.77, - "cuda_time_us": 47.744, - "pct_cuda_time": 0.2063117551938868, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.744, - "pct_cuda_time": 0.2063117551938868, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.105, - "cuda_time_us": 10.24, - "pct_cuda_time": 0.04424917001477465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.24, - "pct_cuda_time": 0.04424917001477465, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 177.987, - "cuda_time_us": 510.97399999999993, - "pct_cuda_time": 2.2080249413212365, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.16, - "cuda_time_us": 317.503, - "pct_cuda_time": 1.3719965065625972, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.767, - "pct_cuda_time": 1.3688160974677854, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.738, - "cuda_time_us": 45.503, - "pct_cuda_time": 0.1966279280451456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.503, - "pct_cuda_time": 0.1966279280451456, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.902, - "cuda_time_us": 147.968, - "pct_cuda_time": 0.6394005067134937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 147.968, - "pct_cuda_time": 0.6394005067134937, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 905.74, - "cuda_time_us": 706.558, - "pct_cuda_time": 3.0531840886034325, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.27, - "cuda_time_us": 10.08, - "pct_cuda_time": 0.04355777673329379, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.08, - "pct_cuda_time": 0.04355777673329379, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 644.967, - "cuda_time_us": 174.048, - "pct_cuda_time": 0.7520976115948729, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.421, - "cuda_time_us": 83.264, - "pct_cuda_time": 0.3598010636826363, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0033186877511080987, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.496, - "pct_cuda_time": 0.35648237593152826, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.807, - "cuda_time_us": 13.344, - "pct_cuda_time": 0.05766219967550322, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.344, - "pct_cuda_time": 0.05766219967550322, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 278.646, - "cuda_time_us": 30.016000000000002, - "pct_cuda_time": 0.1297053796058082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.632, - "pct_cuda_time": 0.024337043508126058, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.752, - "pct_cuda_time": 0.09831612462657742, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.632, - "pct_cuda_time": 0.00705221147110471, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.55, - "cuda_time_us": 47.424, - "pct_cuda_time": 0.20492896863092508, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.424, - "pct_cuda_time": 0.20492896863092508, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.039, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.496, - "pct_cuda_time": 0.04535539926514402, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.333, - "cuda_time_us": 511.93399999999997, - "pct_cuda_time": 2.2121733010101217, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.915, - "cuda_time_us": 318.23999999999995, - "pct_cuda_time": 1.3751812368654182, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.0033186877511080987, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.472, - "pct_cuda_time": 1.3718625491143102, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.922, - "cuda_time_us": 44.831, - "pct_cuda_time": 0.19372407626292604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.831, - "pct_cuda_time": 0.19372407626292604, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.45, - "cuda_time_us": 148.863, - "pct_cuda_time": 0.6432679878817772, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.863, - "pct_cuda_time": 0.6432679878817772, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 908.96, - "cuda_time_us": 706.335, - "pct_cuda_time": 3.0522204592173687, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.581, - "cuda_time_us": 10.208, - "pct_cuda_time": 0.04411089135847848, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.208, - "pct_cuda_time": 0.04411089135847848, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 654.177, - "cuda_time_us": 173.855, - "pct_cuda_time": 0.7512636184490866, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.494, - "cuda_time_us": 83.232, - "pct_cuda_time": 0.3596627850263402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.496, - "pct_cuda_time": 0.35648237593152826, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 203.38, - "cuda_time_us": 13.056, - "pct_cuda_time": 0.05641769176883768, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.056, - "pct_cuda_time": 0.05641769176883768, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 271.97, - "cuda_time_us": 29.728, - "pct_cuda_time": 0.12846087169914266, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.568, - "pct_cuda_time": 0.024060486195533713, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.09886923925176211, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.005531146251846831, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.745, - "cuda_time_us": 47.839, - "pct_cuda_time": 0.20672226995476609, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.839, - "pct_cuda_time": 0.20672226995476609, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.494, - "cuda_time_us": 10.784, - "pct_cuda_time": 0.046599907171809556, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.784, - "pct_cuda_time": 0.046599907171809556, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 169.525, - "cuda_time_us": 511.488, - "pct_cuda_time": 2.2102460422379937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.235, - "cuda_time_us": 316.832, - "pct_cuda_time": 1.3690969759883869, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.096, - "pct_cuda_time": 1.3659165668935749, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.204, - "cuda_time_us": 44.832, - "pct_cuda_time": 0.19372839747093526, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.832, - "pct_cuda_time": 0.19372839747093526, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.891, - "cuda_time_us": 149.824, - "pct_cuda_time": 0.6474206687786717, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.824, - "pct_cuda_time": 0.6474206687786717, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 883.253, - "cuda_time_us": 706.3340000000001, - "pct_cuda_time": 3.0522161380093595, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.138, - "cuda_time_us": 10.911, - "pct_cuda_time": 0.04714870058898498, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.911, - "pct_cuda_time": 0.04714870058898498, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 632.377, - "cuda_time_us": 173.663, - "pct_cuda_time": 0.7504339465113096, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.076, - "cuda_time_us": 83.26299999999999, - "pct_cuda_time": 0.35979674247462706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.767, - "pct_cuda_time": 0.0033143665430988433, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.496, - "pct_cuda_time": 0.35648237593152826, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.028, - "cuda_time_us": 13.504, - "pct_cuda_time": 0.058353592956984064, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.504, - "pct_cuda_time": 0.058353592956984064, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 273.796, - "cuda_time_us": 29.632, - "pct_cuda_time": 0.12804603573025414, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.664, - "pct_cuda_time": 0.02447532216442223, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.656, - "pct_cuda_time": 0.09790128865768892, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.005669424908143003, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.525, - "cuda_time_us": 47.264, - "pct_cuda_time": 0.20423757534944426, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.264, - "pct_cuda_time": 0.20423757534944426, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.747, - "cuda_time_us": 10.016, - "pct_cuda_time": 0.043281219420701456, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.016, - "pct_cuda_time": 0.043281219420701456, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 169.994, - "cuda_time_us": 511.744, - "pct_cuda_time": 2.211352271488363, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.064, - "cuda_time_us": 318.336, - "pct_cuda_time": 1.3755960728343068, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.6, - "pct_cuda_time": 1.372415663739495, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.433, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.19414323343982376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.19414323343982376, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.633, - "cuda_time_us": 148.48, - "pct_cuda_time": 0.6416129652142324, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.48, - "pct_cuda_time": 0.6416129652142324, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 970.826, - "cuda_time_us": 708.479, - "pct_cuda_time": 3.061485129189212, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.501, - "cuda_time_us": 10.111, - "pct_cuda_time": 0.04369173418158071, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.111, - "pct_cuda_time": 0.04369173418158071, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 694.777, - "cuda_time_us": 174.912, - "pct_cuda_time": 0.7558311353148695, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.2, - "cuda_time_us": 84.032, - "pct_cuda_time": 0.3631197514337445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0031847303028211833, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 83.295, - "pct_cuda_time": 0.3599350211309233, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 189.791, - "cuda_time_us": 13.119, - "pct_cuda_time": 0.05668992787342077, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.119, - "pct_cuda_time": 0.05668992787342077, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 318.196, - "cuda_time_us": 29.953, - "pct_cuda_time": 0.1294331435012251, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.761, - "pct_cuda_time": 0.024894479341319997, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.688, - "pct_cuda_time": 0.09803956731398507, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 76.817, - "cuda_time_us": 47.808, - "pct_cuda_time": 0.20658831250647916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.808, - "pct_cuda_time": 0.20658831250647916, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 34.14, - "cuda_time_us": 10.592, - "pct_cuda_time": 0.04577023523403253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.592, - "pct_cuda_time": 0.04577023523403253, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 189.43, - "cuda_time_us": 512.864, - "pct_cuda_time": 2.216192024458729, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.319, - "cuda_time_us": 318.91200000000003, - "pct_cuda_time": 1.3780850886476381, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0031847303028211833, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 318.175, - "pct_cuda_time": 1.374900358344817, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 40.092, - "cuda_time_us": 45.344, - "pct_cuda_time": 0.19594085597167402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.344, - "pct_cuda_time": 0.19594085597167402, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 57.74, - "cuda_time_us": 148.608, - "pct_cuda_time": 0.6421660798394171, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.608, - "pct_cuda_time": 0.6421660798394171, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 903.507, - "cuda_time_us": 707.551, - "pct_cuda_time": 3.057475048156623, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.209, - "cuda_time_us": 10.464, - "pct_cuda_time": 0.04521712060884785, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.464, - "pct_cuda_time": 0.04521712060884785, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 644.007, - "cuda_time_us": 173.888, - "pct_cuda_time": 0.751406218313392, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.482, - "cuda_time_us": 83.232, - "pct_cuda_time": 0.3596627850263402, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.496, - "pct_cuda_time": 0.35648237593152826, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 180.125, - "cuda_time_us": 13.376, - "pct_cuda_time": 0.05780047833179938, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.376, - "pct_cuda_time": 0.05780047833179938, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 281.433, - "cuda_time_us": 29.696, - "pct_cuda_time": 0.1283225930428465, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.536, - "pct_cuda_time": 0.023922207539237544, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.656, - "pct_cuda_time": 0.09790128865768892, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.668, - "cuda_time_us": 47.584, - "pct_cuda_time": 0.20562036191240596, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.584, - "pct_cuda_time": 0.20562036191240596, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.936, - "cuda_time_us": 10.336, - "pct_cuda_time": 0.044664005983663164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.336, - "pct_cuda_time": 0.044664005983663164, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.057, - "cuda_time_us": 512.863, - "pct_cuda_time": 2.21618770325072, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.822, - "cuda_time_us": 318.36699999999996, - "pct_cuda_time": 1.3757300302825937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.631, - "pct_cuda_time": 1.3725496211877817, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.199, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.19345184015834294, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.768, - "pct_cuda_time": 0.19345184015834294, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.175, - "cuda_time_us": 149.728, - "pct_cuda_time": 0.6470058328097831, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.728, - "pct_cuda_time": 0.6470058328097831, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 932.101, - "cuda_time_us": 707.425, - "pct_cuda_time": 3.0569305759474563, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.961, - "cuda_time_us": 10.208, - "pct_cuda_time": 0.04411089135847848, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.208, - "pct_cuda_time": 0.04411089135847848, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 671.549, - "cuda_time_us": 174.368, - "pct_cuda_time": 0.7534803981578346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 63.168, - "cuda_time_us": 83.84, - "pct_cuda_time": 0.3622900794959675, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 83.104, - "pct_cuda_time": 0.3591096704011555, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 190.744, - "cuda_time_us": 12.896, - "pct_cuda_time": 0.05572629848735683, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.896, - "pct_cuda_time": 0.05572629848735683, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 294.992, - "cuda_time_us": 29.471999999999998, - "pct_cuda_time": 0.12735464244877326, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.568, - "pct_cuda_time": 0.024060486195533713, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.592, - "pct_cuda_time": 0.09762473134509657, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.005669424908143003, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.336, - "cuda_time_us": 48.16, - "pct_cuda_time": 0.208109377725737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 48.16, - "pct_cuda_time": 0.208109377725737, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.09, - "cuda_time_us": 10.625, - "pct_cuda_time": 0.045912835098337955, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.625, - "pct_cuda_time": 0.045912835098337955, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.12, - "cuda_time_us": 512.2239999999999, - "pct_cuda_time": 2.2134264513328055, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.825, - "cuda_time_us": 317.792, - "pct_cuda_time": 1.373245335677272, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.056, - "pct_cuda_time": 1.37006492658246, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.75, - "cuda_time_us": 45.536, - "pct_cuda_time": 0.19677052790945102, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.536, - "pct_cuda_time": 0.19677052790945102, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.845, - "cuda_time_us": 148.896, - "pct_cuda_time": 0.6434105877460826, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.896, - "pct_cuda_time": 0.6434105877460826, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 891.98, - "cuda_time_us": 707.7739999999999, - "pct_cuda_time": 3.058438677542686, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.167, - "cuda_time_us": 10.433, - "pct_cuda_time": 0.04508316316056093, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.433, - "pct_cuda_time": 0.04508316316056093, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 635.597, - "cuda_time_us": 174.015, - "pct_cuda_time": 0.7519550117305673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.312, - "cuda_time_us": 83.327, - "pct_cuda_time": 0.3600732997872194, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0031760878868026723, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.592, - "pct_cuda_time": 0.3568972119004168, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 179.761, - "cuda_time_us": 13.408, - "pct_cuda_time": 0.05793875698809556, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.408, - "pct_cuda_time": 0.05793875698809556, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 273.137, - "cuda_time_us": 29.823999999999998, - "pct_cuda_time": 0.12887570766803116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.6, - "pct_cuda_time": 0.024198764851829885, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.09886923925176211, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.005807703564439173, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 72.799, - "cuda_time_us": 47.456, - "pct_cuda_time": 0.20506724728722125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.456, - "pct_cuda_time": 0.20506724728722125, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.183, - "cuda_time_us": 9.92, - "pct_cuda_time": 0.04286638345181294, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.92, - "pct_cuda_time": 0.04286638345181294, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.961, - "cuda_time_us": 513.406, - "pct_cuda_time": 2.218534119199745, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.65, - "cuda_time_us": 318.239, - "pct_cuda_time": 1.3751769156574092, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.503, - "pct_cuda_time": 1.3719965065625972, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.943, - "cuda_time_us": 45.504, - "pct_cuda_time": 0.19663224925315484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.504, - "pct_cuda_time": 0.19663224925315484, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.204, - "cuda_time_us": 149.663, - "pct_cuda_time": 0.6467249542891815, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.663, - "pct_cuda_time": 0.6467249542891815, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 914.511, - "cuda_time_us": 706.014, - "pct_cuda_time": 3.0508333514463972, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.581, - "cuda_time_us": 10.335, - "pct_cuda_time": 0.044659684775653914, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.335, - "pct_cuda_time": 0.044659684775653914, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 642.234, - "cuda_time_us": 174.336, - "pct_cuda_time": 0.7533421195015384, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.377, - "cuda_time_us": 83.295, - "pct_cuda_time": 0.3599350211309233, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.559, - "pct_cuda_time": 0.35675461203611136, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 179.935, - "cuda_time_us": 13.632, - "pct_cuda_time": 0.05890670758216875, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.632, - "pct_cuda_time": 0.05890670758216875, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 279.742, - "cuda_time_us": 29.921, - "pct_cuda_time": 0.12929486484492894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.537, - "pct_cuda_time": 0.0239265287472468, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.09886923925176211, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 69.543, - "cuda_time_us": 47.488, - "pct_cuda_time": 0.20520552594351743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.488, - "pct_cuda_time": 0.20520552594351743, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 39.964, - "cuda_time_us": 10.272, - "pct_cuda_time": 0.04438744867107082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.272, - "pct_cuda_time": 0.04438744867107082, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 176.458, - "cuda_time_us": 511.071, - "pct_cuda_time": 2.208444098498134, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.327, - "cuda_time_us": 318.271, - "pct_cuda_time": 1.3753151943137054, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.535, - "pct_cuda_time": 1.3721347852188936, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.723, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.1940049547835276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.1940049547835276, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.5, - "cuda_time_us": 147.904, - "pct_cuda_time": 0.6391239494009013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 147.904, - "pct_cuda_time": 0.6391239494009013, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 930.027, - "cuda_time_us": 707.713, - "pct_cuda_time": 3.058175083854122, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.897, - "cuda_time_us": 10.432, - "pct_cuda_time": 0.04507884195255167, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.432, - "pct_cuda_time": 0.04507884195255167, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 657.581, - "cuda_time_us": 174.33800000000002, - "pct_cuda_time": 0.753350761917557, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.446, - "cuda_time_us": 83.32900000000001, - "pct_cuda_time": 0.36008194220323797, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.593, - "pct_cuda_time": 0.3569015331084261, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 200.62, - "cuda_time_us": 13.281, - "pct_cuda_time": 0.05738996357092013, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.281, - "pct_cuda_time": 0.05738996357092013, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 273.62, - "cuda_time_us": 29.567999999999998, - "pct_cuda_time": 0.12776947841766179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.568, - "pct_cuda_time": 0.024060486195533713, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.528, - "pct_cuda_time": 0.09734817403250423, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.0063608181896238555, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.416, - "cuda_time_us": 48.16, - "pct_cuda_time": 0.208109377725737, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 48.16, - "pct_cuda_time": 0.208109377725737, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.282, - "cuda_time_us": 10.944, - "pct_cuda_time": 0.04729130045329041, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.944, - "pct_cuda_time": 0.04729130045329041, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 186.32, - "cuda_time_us": 511.99899999999997, - "pct_cuda_time": 2.212454179530723, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 67.673, - "cuda_time_us": 317.631, - "pct_cuda_time": 1.3725496211877817, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.895, - "pct_cuda_time": 1.36936921209297, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.964, - "cuda_time_us": 45.856, - "pct_cuda_time": 0.19815331447241274, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.856, - "pct_cuda_time": 0.19815331447241274, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.996, - "cuda_time_us": 148.512, - "pct_cuda_time": 0.6417512438705286, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.512, - "pct_cuda_time": 0.6417512438705286, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 882.914, - "cuda_time_us": 707.23, - "pct_cuda_time": 3.0560879403856522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.664, - "cuda_time_us": 10.368, - "pct_cuda_time": 0.04480228463995933, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.368, - "pct_cuda_time": 0.04480228463995933, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 625.283, - "cuda_time_us": 173.856, - "pct_cuda_time": 0.7512679396570958, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.514, - "cuda_time_us": 83.775, - "pct_cuda_time": 0.3620092009753659, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 83.039, - "pct_cuda_time": 0.35882879188055394, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 182.0, - "cuda_time_us": 13.376, - "pct_cuda_time": 0.05780047833179938, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.376, - "pct_cuda_time": 0.05780047833179938, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 269.336, - "cuda_time_us": 29.408999999999995, - "pct_cuda_time": 0.12708240634419019, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.536, - "pct_cuda_time": 0.023922207539237544, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.56, - "pct_cuda_time": 0.09748645268880039, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.313, - "pct_cuda_time": 0.005673746116152257, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.544, - "cuda_time_us": 47.296, - "pct_cuda_time": 0.20437585400574043, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.296, - "pct_cuda_time": 0.20437585400574043, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.138, - "cuda_time_us": 10.72, - "pct_cuda_time": 0.04632334985921721, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.72, - "pct_cuda_time": 0.04632334985921721, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.649, - "cuda_time_us": 512.2860000000001, - "pct_cuda_time": 2.21369436622938, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.672, - "cuda_time_us": 317.791, - "pct_cuda_time": 1.3732410144692628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.055, - "pct_cuda_time": 1.3700606053744508, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.998, - "cuda_time_us": 45.791, - "pct_cuda_time": 0.19787243595181114, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.791, - "pct_cuda_time": 0.19787243595181114, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 60.769, - "cuda_time_us": 148.704, - "pct_cuda_time": 0.6425809158083057, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.704, - "pct_cuda_time": 0.6425809158083057, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 944.628, - "cuda_time_us": 706.081, - "pct_cuda_time": 3.0511228723830177, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.385, - "cuda_time_us": 10.656, - "pct_cuda_time": 0.04604679254662487, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.656, - "pct_cuda_time": 0.04604679254662487, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 674.3, - "cuda_time_us": 173.793, - "pct_cuda_time": 0.7509957035525128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.749, - "cuda_time_us": 83.68, - "pct_cuda_time": 0.3615986862144866, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.944, - "pct_cuda_time": 0.35841827711967467, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 188.565, - "cuda_time_us": 12.992, - "pct_cuda_time": 0.056141134456245335, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.992, - "pct_cuda_time": 0.056141134456245335, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 300.634, - "cuda_time_us": 29.632999999999996, - "pct_cuda_time": 0.1280503569382634, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.568, - "pct_cuda_time": 0.024060486195533713, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.784, - "pct_cuda_time": 0.0984544032828736, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.281, - "pct_cuda_time": 0.005535467459856086, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 68.445, - "cuda_time_us": 47.488, - "pct_cuda_time": 0.20520552594351743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.488, - "pct_cuda_time": 0.20520552594351743, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 35.134, - "cuda_time_us": 10.304, - "pct_cuda_time": 0.044525727327366996, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.304, - "pct_cuda_time": 0.044525727327366996, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 182.862, - "cuda_time_us": 511.32800000000003, - "pct_cuda_time": 2.209554648956513, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.173, - "cuda_time_us": 318.144, - "pct_cuda_time": 1.37476640089653, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.408, - "pct_cuda_time": 1.371585991801718, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.234, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.1940049547835276, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.896, - "pct_cuda_time": 0.1940049547835276, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 62.103, - "cuda_time_us": 148.288, - "pct_cuda_time": 0.6407832932764554, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.288, - "pct_cuda_time": 0.6407832932764554, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 894.91, - "cuda_time_us": 706.6859999999999, - "pct_cuda_time": 3.0537372032286165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.556, - "cuda_time_us": 10.528, - "pct_cuda_time": 0.04549367792144019, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.528, - "pct_cuda_time": 0.04549367792144019, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 637.228, - "cuda_time_us": 174.239, - "pct_cuda_time": 0.7529229623246407, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.869, - "cuda_time_us": 83.968, - "pct_cuda_time": 0.3628431941211522, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 83.232, - "pct_cuda_time": 0.3596627850263402, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 180.642, - "cuda_time_us": 12.8, - "pct_cuda_time": 0.055311462518468316, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 12.8, - "pct_cuda_time": 0.055311462518468316, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 280.074, - "cuda_time_us": 30.112, - "pct_cuda_time": 0.1301202155746967, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.632, - "pct_cuda_time": 0.024337043508126058, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.976, - "pct_cuda_time": 0.09928407522065061, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.963, - "cuda_time_us": 47.359, - "pct_cuda_time": 0.20464809011032353, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.359, - "pct_cuda_time": 0.20464809011032353, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.508, - "cuda_time_us": 10.369, - "pct_cuda_time": 0.04480660584796859, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.369, - "pct_cuda_time": 0.04480660584796859, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 172.029, - "cuda_time_us": 511.54999999999995, - "pct_cuda_time": 2.2105139571345673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.765, - "cuda_time_us": 317.087, - "pct_cuda_time": 1.3701988840307469, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0031760878868026723, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 316.352, - "pct_cuda_time": 1.3670227961439443, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.929, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.19414323343982376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.19414323343982376, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.392, - "cuda_time_us": 149.535, - "pct_cuda_time": 0.6461718396639968, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.535, - "pct_cuda_time": 0.6461718396639968, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 908.767, - "cuda_time_us": 707.549, - "pct_cuda_time": 3.0574664057406045, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.28, - "cuda_time_us": 10.4, - "pct_cuda_time": 0.0449405632962555, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.4, - "pct_cuda_time": 0.0449405632962555, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 642.908, - "cuda_time_us": 174.431, - "pct_cuda_time": 0.7537526342624178, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 64.869, - "cuda_time_us": 83.551, - "pct_cuda_time": 0.36104125038129264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.815, - "pct_cuda_time": 0.3578608412864807, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 176.087, - "cuda_time_us": 13.28, - "pct_cuda_time": 0.057385642362910876, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.28, - "pct_cuda_time": 0.057385642362910876, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 278.893, - "cuda_time_us": 30.048, - "pct_cuda_time": 0.12984365826210434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.728, - "pct_cuda_time": 0.02475187947701457, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.816, - "pct_cuda_time": 0.09859268193916976, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.006499096845920027, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 69.869, - "cuda_time_us": 47.552, - "pct_cuda_time": 0.20548208325610978, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.552, - "pct_cuda_time": 0.20548208325610978, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 33.196, - "cuda_time_us": 10.592, - "pct_cuda_time": 0.04577023523403253, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.592, - "pct_cuda_time": 0.04577023523403253, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.716, - "cuda_time_us": 512.126, - "pct_cuda_time": 2.2130029729478986, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.029, - "cuda_time_us": 317.854, - "pct_cuda_time": 1.3735132505738459, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.767, - "pct_cuda_time": 0.0033143665430988433, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.087, - "pct_cuda_time": 1.3701988840307469, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.804, - "cuda_time_us": 45.92, - "pct_cuda_time": 0.1984298717850051, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.92, - "pct_cuda_time": 0.1984298717850051, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.54, - "cuda_time_us": 148.352, - "pct_cuda_time": 0.6410598505890478, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 148.352, - "pct_cuda_time": 0.6410598505890478, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 917.964, - "cuda_time_us": 709.151, - "pct_cuda_time": 3.0643889809714313, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.235, - "cuda_time_us": 10.689, - "pct_cuda_time": 0.0461893924109303, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.689, - "pct_cuda_time": 0.0461893924109303, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 660.602, - "cuda_time_us": 174.879, - "pct_cuda_time": 0.755688535450564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.85, - "cuda_time_us": 83.422, - "pct_cuda_time": 0.3604838145480987, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0031760878868026723, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.687, - "pct_cuda_time": 0.35730772666129607, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 196.42, - "cuda_time_us": 13.249, - "pct_cuda_time": 0.05725168491462396, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.249, - "pct_cuda_time": 0.05725168491462396, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 277.869, - "cuda_time_us": 29.663999999999998, - "pct_cuda_time": 0.1281843143865503, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.632, - "pct_cuda_time": 0.024337043508126058, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.72, - "pct_cuda_time": 0.09817784597028124, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.005669424908143003, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.441, - "cuda_time_us": 48.544, - "pct_cuda_time": 0.20976872160129106, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 48.544, - "pct_cuda_time": 0.20976872160129106, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.877, - "cuda_time_us": 10.976, - "pct_cuda_time": 0.04742957910958658, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.976, - "pct_cuda_time": 0.04742957910958658, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.385, - "cuda_time_us": 512.607, - "pct_cuda_time": 2.2150814740003506, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.911, - "cuda_time_us": 317.983, - "pct_cuda_time": 1.3740706864070398, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.247, - "pct_cuda_time": 1.370890277312228, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.586, - "cuda_time_us": 45.12, - "pct_cuda_time": 0.19497290537760079, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.12, - "pct_cuda_time": 0.19497290537760079, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.266, - "cuda_time_us": 149.504, - "pct_cuda_time": 0.6460378822157099, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.504, - "pct_cuda_time": 0.6460378822157099, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 883.574, - "cuda_time_us": 707.196, - "pct_cuda_time": 3.0559410193133374, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.793, - "cuda_time_us": 10.528, - "pct_cuda_time": 0.04549367792144019, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.528, - "pct_cuda_time": 0.04549367792144019, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 635.201, - "cuda_time_us": 174.175, - "pct_cuda_time": 0.7526464050120484, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 55.491, - "cuda_time_us": 83.39200000000001, - "pct_cuda_time": 0.36035417830782107, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.656, - "pct_cuda_time": 0.3571737692130092, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 190.718, - "cuda_time_us": 13.632, - "pct_cuda_time": 0.05890670758216875, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.632, - "pct_cuda_time": 0.05890670758216875, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 270.264, - "cuda_time_us": 29.631000000000004, - "pct_cuda_time": 0.12804171452224491, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.568, - "pct_cuda_time": 0.024060486195533713, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.751, - "pct_cuda_time": 0.09831180341856817, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.005669424908143003, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 64.066, - "cuda_time_us": 47.52, - "pct_cuda_time": 0.2053438045998136, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.52, - "pct_cuda_time": 0.2053438045998136, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.943, - "cuda_time_us": 10.176, - "pct_cuda_time": 0.043972612702182313, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.176, - "pct_cuda_time": 0.043972612702182313, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 166.946, - "cuda_time_us": 512.317, - "pct_cuda_time": 2.213828323677667, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 57.645, - "cuda_time_us": 317.791, - "pct_cuda_time": 1.3732410144692628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.055, - "pct_cuda_time": 1.3700606053744508, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.872, - "cuda_time_us": 45.279, - "pct_cuda_time": 0.19565997745107241, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 45.279, - "pct_cuda_time": 0.19565997745107241, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.495, - "cuda_time_us": 149.247, - "pct_cuda_time": 0.6449273317573313, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.247, - "pct_cuda_time": 0.6449273317573313, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 925.055, - "cuda_time_us": 707.039, - "pct_cuda_time": 3.055262589655884, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.965, - "cuda_time_us": 10.432, - "pct_cuda_time": 0.04507884195255167, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.432, - "pct_cuda_time": 0.04507884195255167, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 659.624, - "cuda_time_us": 174.337, - "pct_cuda_time": 0.7533464407095476, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.879, - "cuda_time_us": 83.61699999999999, - "pct_cuda_time": 0.3613264501099035, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.737, - "pct_cuda_time": 0.0031847303028211833, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.88, - "pct_cuda_time": 0.3581417198070823, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[1024, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 187.861, - "cuda_time_us": 13.408, - "pct_cuda_time": 0.05793875698809556, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 13.408, - "pct_cuda_time": 0.05793875698809556, - "trace": "_C::rotary_embedding(int64[1024], bfloat16[1024, 4096], bfloat16[1024, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 282.191, - "cuda_time_us": 29.887999999999998, - "pct_cuda_time": 0.12915226498062352, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 5.536, - "pct_cuda_time": 0.023922207539237544, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[1024], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, false, false, true, true, true, false, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, cutlass::bfloat16_t, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, false, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 22.88, - "pct_cuda_time": 0.09886923925176211, - "trace": "_vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.0063608181896238555, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], None, None, bfloat16[1024, 32, 128], int32[3], int32[3], None, None, None, 512, 512, None, None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[1024, 32, 128], bfloat16[1024, 8, 128], bfloat16[1024, 8, 128], bfloat16[1024, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 73.65, - "cuda_time_us": 47.424, - "pct_cuda_time": 0.20492896863092508, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x256x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 47.424, - "pct_cuda_time": 0.20492896863092508, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[1024, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 34.279, - "cuda_time_us": 10.272, - "pct_cuda_time": 0.04438744867107082, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.272, - "pct_cuda_time": 0.04438744867107082, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 177.646, - "cuda_time_us": 511.998, - "pct_cuda_time": 2.212449858322714, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.839, - "cuda_time_us": 317.791, - "pct_cuda_time": 1.3732410144692628, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.0031804090948119278, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize128x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 317.055, - "pct_cuda_time": 1.3700606053744508, - "trace": "mm(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[1024, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[1024, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.588, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.19414323343982376, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 44.928, - "pct_cuda_time": 0.19414323343982376, - "trace": "_C::silu_and_mul(bfloat16[1024, 14336], bfloat16[1024, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.543, - "cuda_time_us": 149.279, - "pct_cuda_time": 0.6450656104136274, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize256x128x64_warpgroupsize2x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 149.279, - "pct_cuda_time": 0.6450656104136274, - "trace": "mm(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[1024, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[1024, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.743, - "cuda_time_us": 10.689, - "pct_cuda_time": 0.0461893924109303, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 10.689, - "pct_cuda_time": 0.0461893924109303, - "trace": "_C::fused_add_rms_norm(bfloat16[1024, 4096], bfloat16[1024, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 157.092, - "cuda_time_us": 356.44800000000004, - "pct_cuda_time": 1.5402859524830466, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.013970465493922505, - "trace": "index_select(bfloat16[1024, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.735, - "pct_cuda_time": 0.0031760878868026723, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 352.48, - "pct_cuda_time": 1.5231393991023212, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 1064.23, - "cuda_time_us": 123.745, - "pct_cuda_time": 0.5347278851053017, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.01396614428591325, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.009541227284435785, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.009541227284435785, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.009541227284435785, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.009679505940731956, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.009541227284435785, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.009541227284435785, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.096, - "pct_cuda_time": 0.017699668005909858, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.608, - "pct_cuda_time": 0.01991212650664859, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 33.696, - "pct_cuda_time": 0.14560742507986782, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.392, - "pct_cuda_time": 0.1183665297895222, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 1.889, - "pct_cuda_time": 0.008162761929483332, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.672, - "pct_cuda_time": 0.020188683819240934, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 27.84, - "pct_cuda_time": 0.12030243097766859, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 3.04, - "pct_cuda_time": 0.013136472348136224, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - }, - "decode_1": { - "metadata": { - "num_running_seqs": 2 - }, - "summary_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cuda_time_us": 6417.871, - "pct_cuda_time": 93.19567841007125, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04832680771376008, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04832680771376008, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cuda_time_us": 6411.406999999999, - "pct_cuda_time": 93.10181287970414, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 204.86199999999997, - "pct_cuda_time": 2.974857716903941, - "invocations": 64 - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 4.288, - "pct_cuda_time": 0.062267233015806264, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 200.57399999999998, - "pct_cuda_time": 2.9125904838881356, - "invocations": 63 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cuda_time_us": 1951.4869999999999, - "pct_cuda_time": 28.33808203272311, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cuda_time_us": 667.1719999999999, - "pct_cuda_time": 9.688188989184116, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 667.1719999999999, - "pct_cuda_time": 9.688188989184116, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cuda_time_us": 117.054, - "pct_cuda_time": 1.699773482610118, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cuda_time_us": 117.054, - "pct_cuda_time": 1.699773482610118, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cuda_time_us": 600.8019999999999, - "pct_cuda_time": 8.724411877416612, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cuda_time_us": 74.87899999999999, - "pct_cuda_time": 1.0873386522832456, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cuda_time_us": 401.59999999999997, - "pct_cuda_time": 5.831744584689318, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cuda_time_us": 79.075, - "pct_cuda_time": 1.1482699278742725, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cuda_time_us": 45.24799999999999, - "pct_cuda_time": 0.6570587125697764, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cuda_time_us": 566.4590000000001, - "pct_cuda_time": 8.225707683512267, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cuda_time_us": 498.2059999999999, - "pct_cuda_time": 7.234587008365851, - "invocations": 32 - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cuda_time_us": 68.25299999999999, - "pct_cuda_time": 0.9911206751464141, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cuda_time_us": 4255.058, - "pct_cuda_time": 61.78887313007709, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cuda_time_us": 2576.089000000001, - "pct_cuda_time": 37.40810028741963, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 2576.089000000001, - "pct_cuda_time": 37.40810028741963, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cuda_time_us": 287.166, - "pct_cuda_time": 4.1700168461327, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cuda_time_us": 287.166, - "pct_cuda_time": 4.1700168461327, - "invocations": 32 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cuda_time_us": 1391.8029999999999, - "pct_cuda_time": 20.210755996524764, - "invocations": 32 - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 1391.8029999999999, - "pct_cuda_time": 20.210755996524764, - "invocations": 32 - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "invocations": 1 - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cuda_time_us": 347.551, - "pct_cuda_time": 5.046884118907761, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010687659398235403, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cuda_time_us": 343.679, - "pct_cuda_time": 4.990657736856175, - "invocations": 1 - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cuda_time_us": 121.02499999999999, - "pct_cuda_time": 1.7574374710209777, - "invocations": 1 - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cuda_time_us": 12.576000000000002, - "pct_cuda_time": 0.18261957145680496, - "invocations": 7 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 4.192, - "pct_cuda_time": 0.06087319048560164, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cuda_time_us": 4.705, - "pct_cuda_time": 0.06832260525638256, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 34.24, - "pct_cuda_time": 0.49720850243964704, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cuda_time_us": 27.968, - "pct_cuda_time": 0.40613105713294534, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cuda_time_us": 1.792, - "pct_cuda_time": 0.026022127230486202, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cuda_time_us": 4.736, - "pct_cuda_time": 0.06877276482342781, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cuda_time_us": 28.32, - "pct_cuda_time": 0.4112425464103623, - "invocations": 1 - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cuda_time_us": 2.496, - "pct_cuda_time": 0.036245105785320066, - "invocations": 1 - }, - "children": [] - } - ] - } - ], - "model_stats": [ - { - "entry": { - "name": "LlamaForCausalLM", - "cpu_time_us": 29745.046, - "cuda_time_us": 6417.871, - "pct_cuda_time": 93.19567841007125, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "VocabParallelEmbedding(weight=bfloat16[128256, 4096])", - "cpu_time_us": 84.102, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04832680771376008, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04832680771376008, - "trace": "index_select(bfloat16[128256, 4096], 0, int64[2]) <- embedding(bfloat16[128256, 4096], int64[2], -1, False, False)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 1359.597, - "cuda_time_us": 207.10500000000002, - "pct_cuda_time": 3.0074289397711187, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 76.13, - "cuda_time_us": 4.288, - "pct_cuda_time": 0.062267233015806264, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rms_norm_kernel(c10::BFloat16*, c10::BFloat16 const*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 4.288, - "pct_cuda_time": 0.062267233015806264, - "trace": "_C::rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 980.707, - "cuda_time_us": 63.903999999999996, - "pct_cuda_time": 0.9279676442728739, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 121.87, - "cuda_time_us": 23.392, - "pct_cuda_time": 0.3396816965265252, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 23.392, - "pct_cuda_time": 0.3396816965265252, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 292.191, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.052044254460972404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.052044254460972404, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 382.249, - "cuda_time_us": 19.104000000000003, - "pct_cuda_time": 0.277414463510719, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03392170156831236, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.8, - "pct_cuda_time": 0.1858723373606157, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 92.76, - "cuda_time_us": 17.823999999999998, - "pct_cuda_time": 0.25882722977465733, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.712, - "pct_cuda_time": 0.22815829411015576, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 43.426, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.045553243929707145, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.045553243929707145, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 217.717, - "cuda_time_us": 135.776, - "pct_cuda_time": 1.9716408185527314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 80.567, - "cuda_time_us": 82.24, - "pct_cuda_time": 1.1942297675419558, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.24, - "pct_cuda_time": 1.1942297675419558, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 48.85, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13057531699583255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13057531699583255, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 61.374, - "cuda_time_us": 44.544, - "pct_cuda_time": 0.6468357340149427, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.544, - "pct_cuda_time": 0.6468357340149427, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 985.424, - "cuda_time_us": 202.367, - "pct_cuda_time": 2.938627132394978, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.715, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 707.163, - "cuda_time_us": 60.576, - "pct_cuda_time": 0.879640836559114, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 54.837, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.3001838248373944, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.672, - "pct_cuda_time": 0.3001838248373944, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 217.906, - "cuda_time_us": 3.903, - "pct_cuda_time": 0.05667654161863149, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.903, - "pct_cuda_time": 0.05667654161863149, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 307.436, - "cuda_time_us": 18.592000000000002, - "pct_cuda_time": 0.26997957001629436, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032992339881509285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.384, - "pct_cuda_time": 0.1798314863963957, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.034851063255115444, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.536, - "pct_cuda_time": 0.022304680483273887, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.368, - "cuda_time_us": 17.409, - "pct_cuda_time": 0.2528009000867937, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.296, - "pct_cuda_time": 0.22211744314593576, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.03068345694085789, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.167, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04693276518355547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04693276518355547, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 183.155, - "cuda_time_us": 135.45499999999998, - "pct_cuda_time": 1.9669794888423593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 61.863, - "cuda_time_us": 82.783, - "pct_cuda_time": 1.2021148206034258, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 82.783, - "pct_cuda_time": 1.2021148206034258, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 39.637, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.1254638277184156, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.1254638277184156, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.939, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.6394008405205179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.6394008405205179, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 919.383, - "cuda_time_us": 200.66899999999998, - "pct_cuda_time": 2.9139700051419837, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.845, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 657.926, - "cuda_time_us": 61.727, - "pct_cuda_time": 0.8963548256452128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.555, - "cuda_time_us": 21.76, - "pct_cuda_time": 0.3159829735130467, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.76, - "pct_cuda_time": 0.3159829735130467, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 200.654, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 291.362, - "cuda_time_us": 18.719, - "pct_cuda_time": 0.2718237721135442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.431, - "pct_cuda_time": 0.03530122282216069, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.18169020977000186, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.221, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.2560391447142481, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.22537020904974653, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.216, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04786212687035854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04786212687035854, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.108, - "cuda_time_us": 132.51, - "pct_cuda_time": 1.9242143299730614, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.447, - "cuda_time_us": 80.319, - "pct_cuda_time": 1.1663343956615073, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.319, - "pct_cuda_time": 1.1663343956615073, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.096, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.62, - "cuda_time_us": 43.455, - "pct_cuda_time": 0.631022064062934, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.455, - "pct_cuda_time": 0.631022064062934, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 934.051, - "cuda_time_us": 199.553, - "pct_cuda_time": 2.897764260728355, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.197, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 665.222, - "cuda_time_us": 60.193, - "pct_cuda_time": 0.8740791877146517, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.173, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.29600169724678055, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.29600169724678055, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 196.145, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05343829699117702, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05343829699117702, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 294.791, - "cuda_time_us": 18.753, - "pct_cuda_time": 0.2723174955096583, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.337, - "pct_cuda_time": 0.03393622284466867, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.704, - "pct_cuda_time": 0.18447829483041112, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03531574409851698, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.01858723373606157, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.008, - "cuda_time_us": 17.375999999999998, - "pct_cuda_time": 0.2523216979670358, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.264, - "pct_cuda_time": 0.2216527623025342, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.834, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.272, - "cuda_time_us": 132.96, - "pct_cuda_time": 1.9307489043333959, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 65.779, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1742484912756899, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1742484912756899, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.738, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12964595530902948, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12964595530902948, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.098, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6268544577486765, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.168, - "pct_cuda_time": 0.6268544577486765, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 904.623, - "cuda_time_us": 200.736, - "pct_cuda_time": 2.914942930657856, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.741, - "cuda_time_us": 3.263, - "pct_cuda_time": 0.047382924750600706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.263, - "pct_cuda_time": 0.047382924750600706, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 644.313, - "cuda_time_us": 60.705, - "pct_cuda_time": 0.8815140812090764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.57, - "cuda_time_us": 20.48, - "pct_cuda_time": 0.29739573977698514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.48, - "pct_cuda_time": 0.29739573977698514, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 181.645, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05343829699117702, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05343829699117702, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 289.436, - "cuda_time_us": 19.265, - "pct_cuda_time": 0.27975238900408295, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03392170156831236, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.864, - "pct_cuda_time": 0.1868016990474188, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03717446747212314, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.021854520916228644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 66.653, - "cuda_time_us": 17.28, - "pct_cuda_time": 0.2509276554368312, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.168, - "pct_cuda_time": 0.22025871977232964, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.621, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.604, - "cuda_time_us": 133.6, - "pct_cuda_time": 1.9400425212014263, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.548, - "cuda_time_us": 81.184, - "pct_cuda_time": 1.178895299709705, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.184, - "pct_cuda_time": 1.178895299709705, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.323, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.336, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6342893512431012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6342893512431012, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 919.888, - "cuda_time_us": 199.935, - "pct_cuda_time": 2.903311388296461, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.374, - "cuda_time_us": 3.072, - "pct_cuda_time": 0.04460936096654777, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.072, - "pct_cuda_time": 0.04460936096654777, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 650.847, - "cuda_time_us": 61.791, - "pct_cuda_time": 0.897284187332016, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.069, - "cuda_time_us": 21.377, - "pct_cuda_time": 0.3104213246685845, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.377, - "pct_cuda_time": 0.3104213246685845, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 178.792, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.054367658677980095, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.744, - "pct_cuda_time": 0.054367658677980095, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 289.776, - "cuda_time_us": 19.200000000000003, - "pct_cuda_time": 0.2788085060409236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.18169020977000186, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03717446747212314, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.664, - "pct_cuda_time": 0.02416340385688004, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 61.072, - "cuda_time_us": 17.47, - "pct_cuda_time": 0.25368669794452786, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.327, - "pct_cuda_time": 0.22256760271298104, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.143, - "pct_cuda_time": 0.031119095231546833, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.988, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 173.303, - "cuda_time_us": 131.904, - "pct_cuda_time": 1.915414436501145, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.537, - "cuda_time_us": 79.808, - "pct_cuda_time": 1.158914023443439, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.808, - "pct_cuda_time": 1.158914023443439, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.302, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.1254638277184156, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.64, - "pct_cuda_time": 0.1254638277184156, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.423, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.6310365853392904, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.456, - "pct_cuda_time": 0.6310365853392904, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 893.376, - "cuda_time_us": 199.137, - "pct_cuda_time": 2.8917234097641353, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.411, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 630.705, - "cuda_time_us": 60.513, - "pct_cuda_time": 0.878725996148667, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.275, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.301577867367599, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.301577867367599, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 171.817, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.05249441402801764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.05249441402801764, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 292.513, - "cuda_time_us": 18.498, - "pct_cuda_time": 0.26861457003880235, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032992339881509285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.449, - "pct_cuda_time": 0.18077536935955507, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.465, - "pct_cuda_time": 0.03579494621827482, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 61.641, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.2560391447142481, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.22537020904974653, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.708, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.628, - "cuda_time_us": 132.28799999999998, - "pct_cuda_time": 1.9209906066219633, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 64.702, - "cuda_time_us": 80.192, - "pct_cuda_time": 1.1644901935642573, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.192, - "pct_cuda_time": 1.1644901935642573, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.321, - "cuda_time_us": 8.96, - "pct_cuda_time": 0.13011063615243101, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.96, - "pct_cuda_time": 0.13011063615243101, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 56.36, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.626389776905275, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.136, - "pct_cuda_time": 0.626389776905275, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 899.671, - "cuda_time_us": 198.04700000000003, - "pct_cuda_time": 2.8758952185357707, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.704, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 646.019, - "cuda_time_us": 60.127, - "pct_cuda_time": 0.873120783475136, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.442, - "cuda_time_us": 20.48, - "pct_cuda_time": 0.29739573977698514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.48, - "pct_cuda_time": 0.29739573977698514, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 194.513, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 286.038, - "cuda_time_us": 18.656000000000002, - "pct_cuda_time": 0.27090893170309743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03392170156831236, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.1830842523002065, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.034851063255115444, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.292, - "cuda_time_us": 17.375, - "pct_cuda_time": 0.25230717669067954, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.264, - "pct_cuda_time": 0.2216527623025342, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.0306544143881453, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.982, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 169.489, - "cuda_time_us": 131.68, - "pct_cuda_time": 1.9121616705973343, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.089, - "cuda_time_us": 79.52, - "pct_cuda_time": 1.154731895852825, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.52, - "pct_cuda_time": 1.154731895852825, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.157, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.41, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.6305719044958887, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.424, - "pct_cuda_time": 0.6305719044958887, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 865.314, - "cuda_time_us": 199.67900000000003, - "pct_cuda_time": 2.899593941549249, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.607, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 609.845, - "cuda_time_us": 60.928000000000004, - "pct_cuda_time": 0.8847523258365307, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.951, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.30064850568079593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.30064850568079593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 180.087, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05343829699117702, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05343829699117702, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 270.454, - "cuda_time_us": 18.944000000000003, - "pct_cuda_time": 0.2750910592937113, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03345702072491082, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.704, - "pct_cuda_time": 0.18447829483041112, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021375318796470807, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 57.173, - "cuda_time_us": 17.6, - "pct_cuda_time": 0.25557446387084665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.456, - "pct_cuda_time": 0.22444084736294345, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.031133616507903132, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 36.25, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04552420137699455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04552420137699455, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 167.191, - "cuda_time_us": 132.512, - "pct_cuda_time": 1.9242433725257742, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.142, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.163096151034053, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.096, - "pct_cuda_time": 1.163096151034053, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 35.753, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.617, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6342893512431012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6342893512431012, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 919.734, - "cuda_time_us": 199.90200000000004, - "pct_cuda_time": 2.902832186176704, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.137, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 654.655, - "cuda_time_us": 60.768, - "pct_cuda_time": 0.8824289216195231, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 57.681, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2992544631505913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2992544631505913, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 193.579, - "cuda_time_us": 3.743, - "pct_cuda_time": 0.0543531374016238, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.743, - "pct_cuda_time": 0.0543531374016238, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 284.997, - "cuda_time_us": 18.849, - "pct_cuda_time": 0.2737115380398629, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.241, - "pct_cuda_time": 0.03254218031446405, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.18169020977000186, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.592, - "pct_cuda_time": 0.03763914831552468, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.006, - "cuda_time_us": 17.567999999999998, - "pct_cuda_time": 0.255109783027445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.22397616651954194, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.031133616507903132, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.03, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04693276518355547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04693276518355547, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.47, - "cuda_time_us": 132.76600000000002, - "pct_cuda_time": 1.9279317767202744, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 68.749, - "cuda_time_us": 80.896, - "pct_cuda_time": 1.1747131721190913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.896, - "pct_cuda_time": 1.1747131721190913, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.255, - "cuda_time_us": 8.671, - "pct_cuda_time": 0.12591398728546083, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.671, - "pct_cuda_time": 0.12591398728546083, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.541, - "cuda_time_us": 43.199, - "pct_cuda_time": 0.6273046173157217, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.199, - "pct_cuda_time": 0.6273046173157217, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 876.567, - "cuda_time_us": 200.001, - "pct_cuda_time": 2.9042697925359766, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.368, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.045553243929707145, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.045553243929707145, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 619.458, - "cuda_time_us": 60.32000000000001, - "pct_cuda_time": 0.8759233898119017, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.655, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29971914399399285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29971914399399285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 180.262, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 276.612, - "cuda_time_us": 18.528000000000002, - "pct_cuda_time": 0.2690502083294913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032992339881509285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.18169020977000186, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03531574409851698, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 57.518, - "cuda_time_us": 17.536, - "pct_cuda_time": 0.25464510218404357, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.456, - "pct_cuda_time": 0.22444084736294345, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.030204254821100056, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.641, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.423, - "cuda_time_us": 133.44, - "pct_cuda_time": 1.9377191169844186, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 57.314, - "cuda_time_us": 80.128, - "pct_cuda_time": 1.1635608318774544, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.128, - "pct_cuda_time": 1.1635608318774544, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.629, - "cuda_time_us": 9.824, - "pct_cuda_time": 0.14265701892427254, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.824, - "pct_cuda_time": 0.14265701892427254, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 61.953, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.6315012661826919, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.6315012661826919, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 867.561, - "cuda_time_us": 199.361, - "pct_cuda_time": 2.8949761756679457, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.021, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 620.648, - "cuda_time_us": 61.057, - "pct_cuda_time": 0.8866255704864932, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.473, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.301577867367599, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.301577867367599, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 171.383, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.05157957361757086, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.552, - "pct_cuda_time": 0.05157957361757086, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 276.061, - "cuda_time_us": 18.688000000000002, - "pct_cuda_time": 0.27137361254649894, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03392170156831236, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.1830842523002065, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03531574409851698, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 57.872, - "cuda_time_us": 18.049, - "pct_cuda_time": 0.26209451695482444, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.936, - "pct_cuda_time": 0.23141106001396658, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.113, - "pct_cuda_time": 0.03068345694085789, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.075, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 165.815, - "cuda_time_us": 132.032, - "pct_cuda_time": 1.9172731598747512, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.287, - "cuda_time_us": 79.456, - "pct_cuda_time": 1.153802534166022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.456, - "pct_cuda_time": 1.153802534166022, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 35.869, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.12732255109202179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.12732255109202179, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.777, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6361480746167073, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.808, - "pct_cuda_time": 0.6361480746167073, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 927.7, - "cuda_time_us": 200.574, - "pct_cuda_time": 2.9125904838881356, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.315, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 662.314, - "cuda_time_us": 60.768, - "pct_cuda_time": 0.8824289216195231, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 59.882, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.301577867367599, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.768, - "pct_cuda_time": 0.301577867367599, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 176.953, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 307.863, - "cuda_time_us": 18.912000000000003, - "pct_cuda_time": 0.27462637845030974, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03345702072491082, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.672, - "pct_cuda_time": 0.18401361398700955, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.472, - "pct_cuda_time": 0.021375318796470807, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.336, - "cuda_time_us": 17.439999999999998, - "pct_cuda_time": 0.2532510596538389, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.22258212398933733, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 32.306, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 179.659, - "cuda_time_us": 133.502, - "pct_cuda_time": 1.9386194361185094, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.821, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1756425338058942, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.96, - "pct_cuda_time": 1.1756425338058942, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.295, - "cuda_time_us": 9.087, - "pct_cuda_time": 0.13195483824968085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.087, - "pct_cuda_time": 0.13195483824968085, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 55.793, - "cuda_time_us": 43.455, - "pct_cuda_time": 0.631022064062934, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.455, - "pct_cuda_time": 0.631022064062934, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 886.506, - "cuda_time_us": 200.8, - "pct_cuda_time": 2.915872292344659, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.122, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04832680771376008, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04832680771376008, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 634.16, - "cuda_time_us": 61.184, - "pct_cuda_time": 0.888469772583743, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.776, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29971914399399285, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.64, - "pct_cuda_time": 0.29971914399399285, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 179.943, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.056691062894987786, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.904, - "pct_cuda_time": 0.056691062894987786, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 281.481, - "cuda_time_us": 19.072000000000003, - "pct_cuda_time": 0.27694978266731746, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032992339881509285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.832, - "pct_cuda_time": 0.18633701820401727, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.069, - "cuda_time_us": 17.567999999999998, - "pct_cuda_time": 0.255109783027445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.456, - "pct_cuda_time": 0.22444084736294345, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.593, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04832680771376008, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.328, - "pct_cuda_time": 0.04832680771376008, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.894, - "cuda_time_us": 132.96, - "pct_cuda_time": 1.9307489043333959, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.068, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1742484912756899, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.864, - "pct_cuda_time": 1.1742484912756899, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.069, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12778723193542332, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12778723193542332, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.457, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.6287131811222826, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.296, - "pct_cuda_time": 0.6287131811222826, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 903.402, - "cuda_time_us": 199.904, - "pct_cuda_time": 2.902861228729416, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.174, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 647.863, - "cuda_time_us": 60.193000000000005, - "pct_cuda_time": 0.8740791877146518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.533, - "cuda_time_us": 20.832, - "pct_cuda_time": 0.3025072290544021, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.832, - "pct_cuda_time": 0.3025072290544021, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 191.394, - "cuda_time_us": 3.585, - "pct_cuda_time": 0.0520587757373287, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.585, - "pct_cuda_time": 0.0520587757373287, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 289.919, - "cuda_time_us": 18.368000000000002, - "pct_cuda_time": 0.26672680411248356, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.032527659038107753, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.384, - "pct_cuda_time": 0.1798314863963957, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03531574409851698, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.88, - "cuda_time_us": 17.408, - "pct_cuda_time": 0.25278637881043736, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.328, - "pct_cuda_time": 0.22258212398933733, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.030204254821100056, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.681, - "cuda_time_us": 3.199, - "pct_cuda_time": 0.04645356306379763, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.199, - "pct_cuda_time": 0.04645356306379763, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.695, - "cuda_time_us": 133.344, - "pct_cuda_time": 1.9363250744542142, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.577, - "cuda_time_us": 80.64, - "pct_cuda_time": 1.170995725371879, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.64, - "pct_cuda_time": 1.170995725371879, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.343, - "cuda_time_us": 9.024, - "pct_cuda_time": 0.13103999783923406, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.024, - "pct_cuda_time": 0.13103999783923406, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.234, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6342893512431012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6342893512431012, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 873.273, - "cuda_time_us": 200.18900000000002, - "pct_cuda_time": 2.906999792490961, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.965, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 627.294, - "cuda_time_us": 61.24700000000001, - "pct_cuda_time": 0.8893846129941899, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.865, - "cuda_time_us": 21.344, - "pct_cuda_time": 0.3099421225488267, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.344, - "pct_cuda_time": 0.3099421225488267, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 187.873, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 278.059, - "cuda_time_us": 18.687, - "pct_cuda_time": 0.2713590912701427, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.527, - "pct_cuda_time": 0.03669526535236531, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.384, - "pct_cuda_time": 0.1798314863963957, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03531574409851698, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.01951659542286465, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.248, - "cuda_time_us": 17.6, - "pct_cuda_time": 0.25557446387084665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.488, - "pct_cuda_time": 0.22490552820634502, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.59, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 165.261, - "cuda_time_us": 132.638, - "pct_cuda_time": 1.9260730533466677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 57.34, - "cuda_time_us": 80.255, - "pct_cuda_time": 1.1654050339747042, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.255, - "pct_cuda_time": 1.1654050339747042, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.227, - "cuda_time_us": 9.28, - "pct_cuda_time": 0.1347574445864464, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.28, - "pct_cuda_time": 0.1347574445864464, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.859, - "cuda_time_us": 43.103, - "pct_cuda_time": 0.6259105747855171, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.103, - "pct_cuda_time": 0.6259105747855171, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 855.02, - "cuda_time_us": 200.67, - "pct_cuda_time": 2.91398452641834, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.633, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 612.272, - "cuda_time_us": 60.797, - "pct_cuda_time": 0.8828500386338556, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 49.957, - "cuda_time_us": 20.48, - "pct_cuda_time": 0.29739573977698514, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.48, - "pct_cuda_time": 0.29739573977698514, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 171.401, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 283.24, - "cuda_time_us": 18.942, - "pct_cuda_time": 0.27506201674099867, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.479, - "pct_cuda_time": 0.181211007650244, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.036245105785320066, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.503, - "pct_cuda_time": 0.021825478363516046, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 57.07, - "cuda_time_us": 17.727, - "pct_cuda_time": 0.2574186659680965, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.424, - "pct_cuda_time": 0.22397616651954194, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.303, - "pct_cuda_time": 0.03344249944855453, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.724, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.046947286459911766, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.233, - "pct_cuda_time": 0.046947286459911766, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 163.363, - "cuda_time_us": 133.504, - "pct_cuda_time": 1.938648478671222, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 56.679, - "cuda_time_us": 81.088, - "pct_cuda_time": 1.1775012571795005, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.088, - "pct_cuda_time": 1.1775012571795005, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 35.495, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12964595530902948, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.928, - "pct_cuda_time": 0.12964595530902948, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.434, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.6315012661826919, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.488, - "pct_cuda_time": 0.6315012661826919, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 879.631, - "cuda_time_us": 199.742, - "pct_cuda_time": 2.9005087819596955, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.663, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 622.581, - "cuda_time_us": 61.375, - "pct_cuda_time": 0.8912433363677961, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 49.622, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.30111318652419744, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.30111318652419744, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 171.127, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 285.713, - "cuda_time_us": 18.752000000000002, - "pct_cuda_time": 0.2723029742333021, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032992339881509285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.352, - "pct_cuda_time": 0.17936680555299417, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.624, - "pct_cuda_time": 0.038103829158926225, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 63.78, - "cuda_time_us": 18.239, - "pct_cuda_time": 0.2648535594625211, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.128, - "pct_cuda_time": 0.2341991450743758, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.0306544143881453, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 31.402, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 174.289, - "cuda_time_us": 132.063, - "pct_cuda_time": 1.9177233194417962, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 62.75, - "cuda_time_us": 79.391, - "pct_cuda_time": 1.1528586512028627, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.391, - "pct_cuda_time": 1.1528586512028627, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.381, - "cuda_time_us": 9.6, - "pct_cuda_time": 0.13940425302046178, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.6, - "pct_cuda_time": 0.13940425302046178, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.915, - "cuda_time_us": 43.072, - "pct_cuda_time": 0.625460415218472, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.072, - "pct_cuda_time": 0.625460415218472, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 882.846, - "cuda_time_us": 199.935, - "pct_cuda_time": 2.903311388296461, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.169, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04786212687035854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04786212687035854, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 632.812, - "cuda_time_us": 61.05500000000001, - "pct_cuda_time": 0.8865965279337809, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.273, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.30064850568079593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.30064850568079593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.734, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.05249441402801764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.05249441402801764, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 277.594, - "cuda_time_us": 18.528000000000002, - "pct_cuda_time": 0.2690502083294913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032992339881509285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.48, - "pct_cuda_time": 0.18122552892660032, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03531574409851698, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.01951659542286465, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 65.837, - "cuda_time_us": 18.208, - "pct_cuda_time": 0.26440339989547584, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.128, - "pct_cuda_time": 0.2341991450743758, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.030204254821100056, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.052, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.025, - "cuda_time_us": 132.448, - "pct_cuda_time": 1.9233140108389712, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.615, - "cuda_time_us": 80.063, - "pct_cuda_time": 1.162616948914295, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.063, - "pct_cuda_time": 1.162616948914295, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.352, - "cuda_time_us": 9.409, - "pct_cuda_time": 0.13663068923640886, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.409, - "pct_cuda_time": 0.13663068923640886, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.682, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6240663726882673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6240663726882673, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 882.453, - "cuda_time_us": 199.20299999999997, - "pct_cuda_time": 2.8926818140036508, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.047, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.045553243929707145, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.137, - "pct_cuda_time": 0.045553243929707145, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 628.006, - "cuda_time_us": 61.282000000000004, - "pct_cuda_time": 0.8898928576666604, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.011, - "cuda_time_us": 20.641, - "pct_cuda_time": 0.2997336652703491, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.641, - "pct_cuda_time": 0.2997336652703491, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 174.047, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 293.835, - "cuda_time_us": 18.753000000000004, - "pct_cuda_time": 0.27231749550965834, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.036709786628721604, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.18076084808319878, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.465, - "pct_cuda_time": 0.03579494621827482, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.937, - "cuda_time_us": 18.272, - "pct_cuda_time": 0.2653327615822789, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.16, - "pct_cuda_time": 0.23466382591777732, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.957, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.631, - "cuda_time_us": 131.61599999999999, - "pct_cuda_time": 1.9112323089105308, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 66.823, - "cuda_time_us": 79.392, - "pct_cuda_time": 1.152873172479219, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.392, - "pct_cuda_time": 1.152873172479219, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.986, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13057531699583255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.992, - "pct_cuda_time": 0.13057531699583255, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.225, - "cuda_time_us": 43.232, - "pct_cuda_time": 0.6277838194354796, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.232, - "pct_cuda_time": 0.6277838194354796, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 901.197, - "cuda_time_us": 202.274, - "pct_cuda_time": 2.9372766536938424, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.235, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 632.577, - "cuda_time_us": 61.986000000000004, - "pct_cuda_time": 0.9001158362214943, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.512, - "cuda_time_us": 20.865, - "pct_cuda_time": 0.3029864311741599, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.865, - "pct_cuda_time": 0.3029864311741599, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 183.776, - "cuda_time_us": 3.617, - "pct_cuda_time": 0.05252345658073024, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.617, - "pct_cuda_time": 0.05252345658073024, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 284.135, - "cuda_time_us": 19.008000000000003, - "pct_cuda_time": 0.2760204209805144, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.432, - "pct_cuda_time": 0.03531574409851698, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.18169020977000186, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.56, - "pct_cuda_time": 0.03717446747212314, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 62.882, - "cuda_time_us": 18.496000000000002, - "pct_cuda_time": 0.26858552748608977, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.384, - "pct_cuda_time": 0.23791659182158814, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 185.216, - "cuda_time_us": 133.984, - "pct_cuda_time": 1.9456186913222449, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.217, - "cuda_time_us": 80.736, - "pct_cuda_time": 1.1723897679020836, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.736, - "pct_cuda_time": 1.1723897679020836, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 38.068, - "cuda_time_us": 9.568, - "pct_cuda_time": 0.13893957217706024, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.568, - "pct_cuda_time": 0.13893957217706024, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 63.237, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6342893512431012, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.68, - "pct_cuda_time": 0.6342893512431012, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 868.174, - "cuda_time_us": 201.793, - "pct_cuda_time": 2.930291919766463, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.03, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.05018553108736624, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.456, - "pct_cuda_time": 0.05018553108736624, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 617.605, - "cuda_time_us": 61.952999999999996, - "pct_cuda_time": 0.8996366341017362, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.945, - "cuda_time_us": 20.961, - "pct_cuda_time": 0.3043804737043645, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.961, - "pct_cuda_time": 0.3043804737043645, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 177.613, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05343829699117702, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.68, - "pct_cuda_time": 0.05343829699117702, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 276.636, - "cuda_time_us": 18.816, - "pct_cuda_time": 0.2732323359201051, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.271, - "pct_cuda_time": 0.03297781860515299, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.18169020977000186, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.036709786628721604, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.505, - "pct_cuda_time": 0.021854520916228644, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.707, - "cuda_time_us": 18.496000000000002, - "pct_cuda_time": 0.26858552748608977, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.288, - "pct_cuda_time": 0.2365225492913835, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.032062978194706215, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.865, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047397446026957, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.264, - "pct_cuda_time": 0.047397446026957, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 167.557, - "cuda_time_us": 133.12, - "pct_cuda_time": 1.9330723085504036, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.456, - "cuda_time_us": 80.832, - "pct_cuda_time": 1.173783810432288, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.832, - "pct_cuda_time": 1.173783810432288, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.31, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.736, - "pct_cuda_time": 0.12685787024862022, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.156, - "cuda_time_us": 43.552, - "pct_cuda_time": 0.6324306278694949, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.552, - "pct_cuda_time": 0.6324306278694949, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 853.974, - "cuda_time_us": 201.245, - "pct_cuda_time": 2.9223342603232116, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.085, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 607.043, - "cuda_time_us": 60.767, - "pct_cuda_time": 0.8824144003431669, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 58.822, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.30064850568079593, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.704, - "pct_cuda_time": 0.30064850568079593, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 169.975, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.052044254460972404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.052044254460972404, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 268.908, - "cuda_time_us": 18.848000000000003, - "pct_cuda_time": 0.27369701676350666, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.304, - "pct_cuda_time": 0.03345702072491082, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.736, - "pct_cuda_time": 0.18494297567381265, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.344, - "pct_cuda_time": 0.01951659542286465, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.78, - "cuda_time_us": 17.631, - "pct_cuda_time": 0.25602462343789184, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.22537020904974653, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.111, - "pct_cuda_time": 0.0306544143881453, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.956, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04552420137699455, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.135, - "pct_cuda_time": 0.04552420137699455, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 166.665, - "cuda_time_us": 134.175, - "pct_cuda_time": 1.948392255106298, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 57.752, - "cuda_time_us": 80.895, - "pct_cuda_time": 1.174698650842735, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.895, - "pct_cuda_time": 1.174698650842735, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 35.859, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.12732255109202179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.768, - "pct_cuda_time": 0.12732255109202179, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.177, - "cuda_time_us": 44.512, - "pct_cuda_time": 0.6463710531715411, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.512, - "pct_cuda_time": 0.6463710531715411, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 912.356, - "cuda_time_us": 199.325, - "pct_cuda_time": 2.894453409719119, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.855, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 655.94, - "cuda_time_us": 60.352000000000004, - "pct_cuda_time": 0.8763880706553031, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 56.257, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.29600169724678055, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.384, - "pct_cuda_time": 0.29600169724678055, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 196.429, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.616, - "pct_cuda_time": 0.052508935304373935, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 288.274, - "cuda_time_us": 18.784000000000002, - "pct_cuda_time": 0.2727676550767036, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.1830842523002065, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.034851063255115444, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 61.468, - "cuda_time_us": 17.567999999999998, - "pct_cuda_time": 0.255109783027445, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.22304680483273884, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.032062978194706215, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.845, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 171.323, - "cuda_time_us": 132.637, - "pct_cuda_time": 1.9260585320703114, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.678, - "cuda_time_us": 80.607, - "pct_cuda_time": 1.170516523252121, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.607, - "pct_cuda_time": 1.170516523252121, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.85, - "cuda_time_us": 9.087, - "pct_cuda_time": 0.13195483824968085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.087, - "pct_cuda_time": 0.13195483824968085, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.162, - "cuda_time_us": 42.943, - "pct_cuda_time": 0.6235871705685094, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.943, - "pct_cuda_time": 0.6235871705685094, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 872.57, - "cuda_time_us": 199.36, - "pct_cuda_time": 2.89496165439159, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.291, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.104, - "pct_cuda_time": 0.04507404180994931, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 625.06, - "cuda_time_us": 60.577, - "pct_cuda_time": 0.8796553578354701, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.54, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.2987897823071898, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.576, - "pct_cuda_time": 0.2987897823071898, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 184.657, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 277.637, - "cuda_time_us": 18.721, - "pct_cuda_time": 0.27185281466625677, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.305, - "pct_cuda_time": 0.033471542001267125, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.18076084808319878, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 58.494, - "cuda_time_us": 17.631999999999998, - "pct_cuda_time": 0.2560391447142481, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.22537020904974653, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.381, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 166.075, - "cuda_time_us": 132.543, - "pct_cuda_time": 1.9246935320928193, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 59.595, - "cuda_time_us": 80.0, - "pct_cuda_time": 1.1617021085038481, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.0, - "pct_cuda_time": 1.1617021085038481, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.278, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.13522212542984793, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.312, - "pct_cuda_time": 0.13522212542984793, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 51.062, - "cuda_time_us": 43.231, - "pct_cuda_time": 0.6277692981591233, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.231, - "pct_cuda_time": 0.6277692981591233, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 854.864, - "cuda_time_us": 200.575, - "pct_cuda_time": 2.9126050051644916, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.488, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 603.568, - "cuda_time_us": 60.447, - "pct_cuda_time": 0.8777675919091515, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.01, - "cuda_time_us": 20.575, - "pct_cuda_time": 0.29877526103083346, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.575, - "pct_cuda_time": 0.29877526103083346, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 170.825, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 276.482, - "cuda_time_us": 18.753000000000004, - "pct_cuda_time": 0.27231749550965834, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032992339881509285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.512, - "pct_cuda_time": 0.18169020977000186, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.465, - "pct_cuda_time": 0.03579494621827482, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 56.679, - "cuda_time_us": 17.471, - "pct_cuda_time": 0.2537012192208842, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.327, - "pct_cuda_time": 0.22256760271298104, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.031133616507903132, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.036, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04786212687035854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04786212687035854, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 170.402, - "cuda_time_us": 133.696, - "pct_cuda_time": 1.941436563731631, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.969, - "cuda_time_us": 80.8, - "pct_cuda_time": 1.1733191295888865, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.8, - "pct_cuda_time": 1.1733191295888865, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.501, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12871659362222637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12871659362222637, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.938, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.6394008405205179, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.032, - "pct_cuda_time": 0.6394008405205179, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 918.021, - "cuda_time_us": 199.26200000000003, - "pct_cuda_time": 2.8935385693086726, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 26.417, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04879148855716162, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.36, - "pct_cuda_time": 0.04879148855716162, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 661.35, - "cuda_time_us": 61.023, - "pct_cuda_time": 0.8861318470903792, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 50.921, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.3020425482110005, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.3020425482110005, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 186.606, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.05249441402801764, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.615, - "pct_cuda_time": 0.05249441402801764, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 310.95, - "cuda_time_us": 18.496000000000002, - "pct_cuda_time": 0.26858552748608977, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.272, - "pct_cuda_time": 0.032992339881509285, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.18076084808319878, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.355, - "cuda_time_us": 18.112000000000002, - "pct_cuda_time": 0.26300935736527126, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 16.032, - "pct_cuda_time": 0.2328051025441712, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.030204254821100056, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 29.424, - "cuda_time_us": 3.263, - "pct_cuda_time": 0.047382924750600706, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.263, - "pct_cuda_time": 0.047382924750600706, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 168.473, - "cuda_time_us": 131.616, - "pct_cuda_time": 1.9112323089105314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.584, - "cuda_time_us": 79.712, - "pct_cuda_time": 1.1575199809132344, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.712, - "pct_cuda_time": 1.1575199809132344, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.694, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12825191277882486, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.832, - "pct_cuda_time": 0.12825191277882486, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.779, - "cuda_time_us": 43.072, - "pct_cuda_time": 0.625460415218472, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.072, - "pct_cuda_time": 0.625460415218472, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 864.454, - "cuda_time_us": 199.90400000000002, - "pct_cuda_time": 2.9028612287294164, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.26, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 620.11, - "cuda_time_us": 60.193000000000005, - "pct_cuda_time": 0.8740791877146518, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.126, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2992544631505913, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.608, - "pct_cuda_time": 0.2992544631505913, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 176.199, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.052044254460972404, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.584, - "pct_cuda_time": 0.052044254460972404, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 281.815, - "cuda_time_us": 18.657000000000004, - "pct_cuda_time": 0.27092345297945375, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.336, - "pct_cuda_time": 0.03392170156831236, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.1830842523002065, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.433, - "pct_cuda_time": 0.035330265374873285, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.28, - "pct_cuda_time": 0.01858723373606157, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.875, - "cuda_time_us": 17.344, - "pct_cuda_time": 0.25185701712363434, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.232, - "pct_cuda_time": 0.22118808145913268, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.159, - "cuda_time_us": 3.199, - "pct_cuda_time": 0.04645356306379763, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.199, - "pct_cuda_time": 0.04645356306379763, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 165.064, - "cuda_time_us": 133.312, - "pct_cuda_time": 1.9358603936108127, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 58.101, - "cuda_time_us": 80.768, - "pct_cuda_time": 1.172854448745485, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.768, - "pct_cuda_time": 1.172854448745485, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 35.632, - "cuda_time_us": 8.544, - "pct_cuda_time": 0.124069785188211, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.544, - "pct_cuda_time": 0.124069785188211, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.011, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.6389361596771165, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.0, - "pct_cuda_time": 0.6389361596771165, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 886.229, - "cuda_time_us": 202.113, - "pct_cuda_time": 2.9349387282004784, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 24.78, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 622.562, - "cuda_time_us": 61.121, - "pct_cuda_time": 0.8875549321732963, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 49.54, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.30111318652419744, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.736, - "pct_cuda_time": 0.30111318652419744, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 175.677, - "cuda_time_us": 3.681, - "pct_cuda_time": 0.053452818267533314, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.681, - "pct_cuda_time": 0.053452818267533314, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 282.582, - "cuda_time_us": 18.976000000000003, - "pct_cuda_time": 0.2755557401371128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.528, - "pct_cuda_time": 0.036709786628721604, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.448, - "pct_cuda_time": 0.18076084808319878, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.036245105785320066, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 60.172, - "cuda_time_us": 17.728, - "pct_cuda_time": 0.25743318724445274, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.584, - "pct_cuda_time": 0.22629957073654963, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.144, - "pct_cuda_time": 0.031133616507903132, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.377, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 175.558, - "cuda_time_us": 134.72, - "pct_cuda_time": 1.9563063507204805, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 63.807, - "cuda_time_us": 81.568, - "pct_cuda_time": 1.1844714698305236, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 81.568, - "pct_cuda_time": 1.1844714698305236, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.032, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12778723193542332, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.8, - "pct_cuda_time": 0.12778723193542332, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 54.383, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.6440476489545334, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 44.352, - "pct_cuda_time": 0.6440476489545334, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 904.258, - "cuda_time_us": 198.656, - "pct_cuda_time": 2.884738675836756, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.806, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04693276518355547, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.232, - "pct_cuda_time": 0.04693276518355547, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 644.269, - "cuda_time_us": 60.639, - "pct_cuda_time": 0.8805556769695607, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 53.065, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.3020425482110005, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.8, - "pct_cuda_time": 0.3020425482110005, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 192.267, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.648, - "pct_cuda_time": 0.05297361614777548, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 279.72, - "cuda_time_us": 18.719, - "pct_cuda_time": 0.2718237721135442, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.271, - "pct_cuda_time": 0.03297781860515299, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.48, - "pct_cuda_time": 0.18122552892660032, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.464, - "pct_cuda_time": 0.03578042494191853, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.504, - "pct_cuda_time": 0.021839999639872345, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 62.033, - "cuda_time_us": 17.472, - "pct_cuda_time": 0.25371574049724044, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.36, - "pct_cuda_time": 0.22304680483273884, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.112, - "pct_cuda_time": 0.030668935664501594, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.776, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.168, - "pct_cuda_time": 0.04600340349675239, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 178.583, - "cuda_time_us": 131.61700000000002, - "pct_cuda_time": 1.9112468301868877, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 67.661, - "cuda_time_us": 79.712, - "pct_cuda_time": 1.1575199809132344, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.712, - "pct_cuda_time": 1.1575199809132344, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 37.249, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12871659362222637, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.864, - "pct_cuda_time": 0.12871659362222637, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 53.551, - "cuda_time_us": 43.041, - "pct_cuda_time": 0.6250102556514265, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.041, - "pct_cuda_time": 0.6250102556514265, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 898.796, - "cuda_time_us": 199.83800000000002, - "pct_cuda_time": 2.9019028244899006, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.701, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 629.947, - "cuda_time_us": 61.117999999999995, - "pct_cuda_time": 0.8875113683442274, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 51.362, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.3057599949582128, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 21.056, - "pct_cuda_time": 0.3057599949582128, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 181.185, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.053902977834578564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.053902977834578564, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 281.946, - "cuda_time_us": 18.654, - "pct_cuda_time": 0.2708798891503848, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.032527659038107753, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.608, - "pct_cuda_time": 0.1830842523002065, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.463, - "pct_cuda_time": 0.035765903665562225, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.343, - "pct_cuda_time": 0.019502074146508352, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 61.213, - "cuda_time_us": 17.695999999999998, - "pct_cuda_time": 0.25696850640105123, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.296, - "pct_cuda_time": 0.22211744314593576, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.034851063255115444, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 30.944, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 181.115, - "cuda_time_us": 132.38400000000001, - "pct_cuda_time": 1.9223846491521681, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 60.993, - "cuda_time_us": 80.512, - "pct_cuda_time": 1.1691370019982728, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 80.512, - "pct_cuda_time": 1.1691370019982728, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.739, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12918127446562794, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 8.896, - "pct_cuda_time": 0.12918127446562794, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 62.816, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6240663726882673, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 42.976, - "pct_cuda_time": 0.6240663726882673, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "LlamaDecoderLayer", - "cpu_time_us": 888.042, - "cuda_time_us": 199.553, - "pct_cuda_time": 2.897764260728355, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 27.456, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.2, - "pct_cuda_time": 0.046468084340153926, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaAttention", - "cpu_time_us": 633.304, - "cuda_time_us": 60.801, - "pct_cuda_time": 0.882908123739281, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "QKVParallelLinear(weight=bfloat16[6144, 4096])", - "cpu_time_us": 52.16, - "cuda_time_us": 20.929, - "pct_cuda_time": 0.30391579286096293, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 20.929, - "pct_cuda_time": 0.30391579286096293, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 6144]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 6144]) <- linear(bfloat16[2, 4096], bfloat16[6144, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Llama3RotaryEmbedding", - "cpu_time_us": 190.9, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.053902977834578564, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::rotary_embedding_kernel(long const*, c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, int, long, long, int, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.712, - "pct_cuda_time": 0.053902977834578564, - "trace": "_C::rotary_embedding(int64[2], bfloat16[2, 4096], bfloat16[2, 1024], 128, bfloat16[131072, 128], True)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Attention", - "cpu_time_us": 276.519, - "cuda_time_us": 18.560000000000002, - "pct_cuda_time": 0.2695148891728928, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::reshape_and_cache_flash_kernel<__nv_bfloat16, __nv_bfloat16, (vllm::Fp8KVCacheDataType)0>(__nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, __nv_bfloat16*, long const*, int, int, int, int, int, int, float const*, float const*)", - "cpu_time_us": 0, - "cuda_time_us": 2.368, - "pct_cuda_time": 0.034386382411713906, - "trace": "_C_cache_ops::reshape_and_cache_flash(bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], int64[2], None, float32[], float32[]) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > > >(flash::enable_sm90_or_later, cute::C<1>, cute::C<1> >, cute::tuple, cute::C<128>, cute::C<128> >, cutlass::bfloat16_t, float, cutlass::arch::Sm90, true, false, false, true, true, false, true, true, true, true, false>, flash::CollectiveEpilogueFwd, cute::C<128>, cute::C<128> >, cute::tuple, cute::C<1>, cute::C<1> >, float, cutlass::arch::Sm90, 256, true, true, false>, flash::VarlenDynamicPersistentTileScheduler<128, 256, 128, true, true, true> > >::Params)", - "cpu_time_us": 0, - "cuda_time_us": 12.48, - "pct_cuda_time": 0.18122552892660032, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cutlass::device_kernel, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80> >(flash::FlashAttnFwdCombine, cute::C<128> >, 5, 256, 1, false, false, cutlass::bfloat16_t, float, cutlass::arch::Sm80>::Params)", - "cpu_time_us": 0, - "cuda_time_us": 2.4, - "pct_cuda_time": 0.034851063255115444, - "trace": "_vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor, at::detail::Array >(int, at::native::FillFunctor, at::detail::Array)", - "cpu_time_us": 0, - "cuda_time_us": 1.312, - "pct_cuda_time": 0.019051914579463113, - "trace": "fill_(int32[1], 0) <- zero_(int32[1]) <- zeros(None, 3, 0, None, None) <- _vllm_fa3_C::fwd(bfloat16[2, 1, 32, 128], bfloat16[28102, 16, 8, 128], bfloat16[28102, 16, 8, 128], None, None, bfloat16[2, 1, 32, 128], None, None, None, None, int32[2], None, None, int32[2, 33], None, None, None, None, None, None, None, 0.08838834764831845, True, -1, -1, 0, 0.0, True, 0, None, 0) <- vllm::unified_attention_with_output(bfloat16[2, 32, 128], bfloat16[2, 8, 128], bfloat16[2, 8, 128], bfloat16[2, 32, 128], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 4096])", - "cpu_time_us": 59.156, - "cuda_time_us": 17.6, - "pct_cuda_time": 0.25557446387084665, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void cutlass::Kernel2(cutlass_80_tensorop_bf16_s16816gemm_relu_bf16_64x64_64x4_tn_align8::Params)", - "cpu_time_us": 0, - "cuda_time_us": 15.52, - "pct_cuda_time": 0.22537020904974653, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, true, false, false>(cublasLt::cublasSplitKParams, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, float const*, float const*, __nv_bfloat16 const*, __nv_bfloat16 const*, __nv_bfloat16*, void*, long, float*, int*)", - "cpu_time_us": 0, - "cuda_time_us": 2.08, - "pct_cuda_time": 0.030204254821100056, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 4096]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 4096]) <- linear(bfloat16[2, 4096], bfloat16[4096, 4096], None)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 28.611, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04786212687035854, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.296, - "pct_cuda_time": 0.04786212687035854, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "LlamaMLP", - "cpu_time_us": 166.082, - "cuda_time_us": 132.256, - "pct_cuda_time": 1.9205259257785618, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])", - "cpu_time_us": 57.427, - "cuda_time_us": 79.808, - "pct_cuda_time": 1.158914023443439, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 79.808, - "pct_cuda_time": 1.158914023443439, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 28672]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 28672]) <- linear(bfloat16[2, 4096], bfloat16[28672, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "SiluAndMul", - "cpu_time_us": 36.454, - "cuda_time_us": 9.408, - "pct_cuda_time": 0.13661616796005255, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void vllm::act_and_mul_kernel(c10::BFloat16 const&)), true>(c10::BFloat16*, c10::BFloat16 const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 9.408, - "pct_cuda_time": 0.13661616796005255, - "trace": "_C::silu_and_mul(bfloat16[2, 14336], bfloat16[2, 28672])" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "RowParallelLinear(weight=bfloat16[4096, 14336])", - "cpu_time_us": 52.862, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.6249957343750703, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x64x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 43.04, - "pct_cuda_time": 0.6249957343750703, - "trace": "mm(bfloat16[2, 14336], bfloat16[14336, 4096]) <- matmul(bfloat16[2, 14336], bfloat16[14336, 4096]) <- linear(bfloat16[2, 14336], bfloat16[4096, 14336], None)" - }, - "children": [] - } - ] - } - ] - } - ] - }, - { - "entry": { - "name": "RMSNorm(weight=bfloat16[4096])", - "cpu_time_us": 25.686, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "std::enable_if<(((8)>(0)))&&vllm::_typeConvert::exists, void>::type vllm::fused_add_rms_norm_kernel(c10::BFloat16*, c10::BFloat16*, c10::BFloat16 const*, float, int, int)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "_C::fused_add_rms_norm(bfloat16[2, 4096], bfloat16[2, 4096], bfloat16[4096], 1e-05)" - }, - "children": [] - } - ] - } - ] - }, - { - "entry": { - "name": "LogitsProcessor", - "cpu_time_us": 148.71, - "cuda_time_us": 347.551, - "pct_cuda_time": 5.046884118907761, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "void at::native::(anonymous namespace)::indexSelectSmallIndex(at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, at::cuda::detail::TensorInfo, int, int, unsigned int, long)", - "cpu_time_us": 0, - "cuda_time_us": 3.136, - "pct_cuda_time": 0.04553872265335085, - "trace": "index_select(bfloat16[2, 4096], 0, int64[2])" - }, - "children": [] - }, - { - "entry": { - "name": "Memset (Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.736, - "pct_cuda_time": 0.010687659398235403, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - }, - { - "entry": { - "name": "sm90_xmma_gemm_bf16bf16_bf16f32_f32_tn_n_tilesize64x128x64_warpgroupsize1x1x1_execute_segment_k_off_kernel__5x_cublas", - "cpu_time_us": 0, - "cuda_time_us": 343.679, - "pct_cuda_time": 4.990657736856175, - "trace": "mm(bfloat16[2, 4096], bfloat16[4096, 128256]) <- matmul(bfloat16[2, 4096], bfloat16[4096, 128256]) <- linear(bfloat16[2, 4096], bfloat16[128256, 4096], None)" - }, - "children": [] - } - ] - }, - { - "entry": { - "name": "Sampler", - "cpu_time_us": 969.247, - "cuda_time_us": 121.02499999999999, - "pct_cuda_time": 1.7574374710209777, - "trace": "" - }, - "children": [ - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.176, - "pct_cuda_time": 0.03159829735130467, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.032062978194706215, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.24, - "pct_cuda_time": 0.032527659038107753, - "trace": "copy_(int32[2], int32[2], True) <- _to_copy(int32[2], 3, 0, None, None, True, None) <- to(int32[2], 3, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.032062978194706215, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 2.208, - "pct_cuda_time": 0.032062978194706215, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.011152340241636943, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy HtoD (Pinned -> Device)", - "cpu_time_us": 0, - "cuda_time_us": 0.768, - "pct_cuda_time": 0.011152340241636943, - "trace": "copy_(bfloat16[2], bfloat16[2], True) <- _to_copy(bfloat16[2], 15, 0, None, None, True, None) <- to(bfloat16[2], 15, 0, None, None, True, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#7}::operator()() const::{lambda(float)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 4.192, - "pct_cuda_time": 0.06087319048560164, - "trace": "copy_(float32[2, 128256], bfloat16[2, 128256], False) <- _to_copy(bfloat16[2, 128256], 6, None, None, None, False, None) <- to(bfloat16[2, 128256], 6, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::elementwise_kernel<128, 4, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1}>(int, at::native::gpu_kernel_impl > >(at::TensorIteratorBase&, at::native::BinaryFunctor > const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.705, - "pct_cuda_time": 0.06832260525638256, - "trace": "div_(float32[2, 128256], bfloat16[2, 1])" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::SoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 34.24, - "pct_cuda_time": 0.49720850243964704, - "trace": "_softmax(float32[2, 128256], -1, False) <- softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::(anonymous namespace)::cunn_SoftMaxForward<4, float, float, float, at::native::(anonymous namespace)::LogSoftMaxForwardEpilogue>(float*, float const*, int)", - "cpu_time_us": 0, - "cuda_time_us": 27.968, - "pct_cuda_time": 0.40613105713294534, - "trace": "_log_softmax(float32[2, 128256], -1, False) <- log_softmax(float32[2, 128256], -1, 6)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::unrolled_elementwise_kernel, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1> >(int, at::native::direct_copy_kernel_cuda(at::TensorIteratorBase&)::{lambda()#3}::operator()() const::{lambda()#4}::operator()() const::{lambda(long)#1}, at::detail::Array, TrivialOffsetCalculator<1, unsigned int>, TrivialOffsetCalculator<1, unsigned int>, at::native::memory::LoadWithCast<1>, at::native::memory::StoreWithCast<1>)", - "cpu_time_us": 0, - "cuda_time_us": 1.792, - "pct_cuda_time": 0.026022127230486202, - "trace": "copy_(int64[2], int32[2], False) <- _to_copy(int32[2], 4, None, None, None, False, None) <- to(int32[2], 4, False, False, None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::index_elementwise_kernel<128, 4, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1}>(long, at::native::gpu_index_kernel >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1}>(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef, at::native::index_kernel_impl >(at::TensorIteratorBase&, c10::ArrayRef, c10::ArrayRef)::{lambda(char*, char const*, long)#1} const&)::{lambda(int)#1})", - "cpu_time_us": 0, - "cuda_time_us": 4.736, - "pct_cuda_time": 0.06877276482342781, - "trace": "index(float32[2, 128256], None)" - }, - "children": [] - }, - { - "entry": { - "name": "void at::native::reduce_kernel<512, 1, at::native::ReduceOp, unsigned int, long, 4> >(at::native::ReduceOp, unsigned int, long, 4>)", - "cpu_time_us": 0, - "cuda_time_us": 28.32, - "pct_cuda_time": 0.4112425464103623, - "trace": "argmax(float32[2, 128256], -1, False)" - }, - "children": [] - }, - { - "entry": { - "name": "Memcpy DtoH (Device -> Pageable)", - "cpu_time_us": 0, - "cuda_time_us": 2.496, - "pct_cuda_time": 0.036245105785320066, - "trace": "copy_(int64[2], int64[2], False) <- _to_copy(int64[2], 4, 0, None, None, False, None) <- to(int64[2], 4, 0, None, None, False, False, None)" - }, - "children": [] - } - ] - } - ] - } -} \ No newline at end of file diff --git a/Llama3RotaryEmbedding_decode_abs_diff.png b/Llama3RotaryEmbedding_decode_abs_diff.png deleted file mode 100644 index 4adde34701e0b5b717bfea63981e11e7f8c99307..0000000000000000000000000000000000000000 --- a/Llama3RotaryEmbedding_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:983a964b3d2103e4a1de166d55b2679282368f0b8996708d434d7b4836d532d5 -size 46369 diff --git a/Llama3RotaryEmbedding_decode_decode_1_H100.png b/Llama3RotaryEmbedding_decode_decode_1_H100.png deleted file mode 100644 index 9d93e3657c5b853526b8170cfe4f153e55fc7e70..0000000000000000000000000000000000000000 --- a/Llama3RotaryEmbedding_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:d850b57a13d0e0bbf7f7eca3a93eea167b417a1c665e960fb4fcc174b55b628f -size 42067 diff --git a/Llama3RotaryEmbedding_decode_decode_1_MI300x.png b/Llama3RotaryEmbedding_decode_decode_1_MI300x.png deleted file mode 100644 index 966f76be34b2d11f116bfc6b455c4d83d159ae98..0000000000000000000000000000000000000000 --- a/Llama3RotaryEmbedding_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:d9ad049e89b0fb5c014fcce0dd2a330c547558ae2af8ef325ed51ab537ee8b02 -size 42447 diff --git a/Llama3RotaryEmbedding_prefill_abs_diff.png b/Llama3RotaryEmbedding_prefill_abs_diff.png deleted file mode 100644 index 19bd5b73945fea6e41f45f0bf6f8dc3f8d8b216f..0000000000000000000000000000000000000000 --- a/Llama3RotaryEmbedding_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:3cbb8189fda2344c305b206f2a8ddc6bc772ef2bb02720f0f986d3334b4cf008 -size 45958 diff --git a/Llama3RotaryEmbedding_prefill_prefill_H100.png b/Llama3RotaryEmbedding_prefill_prefill_H100.png deleted file mode 100644 index ee705a69a9563da43cb19b63acd78cb55fcf4b7c..0000000000000000000000000000000000000000 --- a/Llama3RotaryEmbedding_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:4518fcbb7d1c61621920b410fb03f88f092c1da4c81b5375b522fd55e78d6f43 -size 42208 diff --git a/Llama3RotaryEmbedding_prefill_prefill_MI300x.png b/Llama3RotaryEmbedding_prefill_prefill_MI300x.png deleted file mode 100644 index 687205472328c09666eed7391200e96bb4ae6302..0000000000000000000000000000000000000000 --- a/Llama3RotaryEmbedding_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:dc22da4fbf9721a91513564098dd41d9e5898060e4e0bb3f0c23567bd67ddc94 -size 44582 diff --git a/LogitsProcessor_decode_abs_diff.png b/LogitsProcessor_decode_abs_diff.png deleted file mode 100644 index bcbe18c773b033c95e2ee215ad74073455f6d49a..0000000000000000000000000000000000000000 --- a/LogitsProcessor_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bef697f6883519e6c046dca3630900cdea0b456575472805bae3100a3607d380 -size 47433 diff --git a/LogitsProcessor_decode_decode_1_H100.png b/LogitsProcessor_decode_decode_1_H100.png deleted file mode 100644 index ee558a8bc326ce30384f664014371545b5e23192..0000000000000000000000000000000000000000 --- a/LogitsProcessor_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:fa947c55808c0bdc6141e3c46af4ecc4e2919cdf62dc60a82ce7101dac53f973 -size 40903 diff --git a/LogitsProcessor_decode_decode_1_MI300x.png b/LogitsProcessor_decode_decode_1_MI300x.png deleted file mode 100644 index b0ec924d56256806cf462eccf7ddd131a53e1a23..0000000000000000000000000000000000000000 --- a/LogitsProcessor_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:283cebef2da9ba8454cb30ba70d6c974006331148ecf88693210504da0a34c07 -size 39323 diff --git a/LogitsProcessor_prefill_abs_diff.png b/LogitsProcessor_prefill_abs_diff.png deleted file mode 100644 index 8def33ee98c2ca0ccdd2015466fdcdf841839a04..0000000000000000000000000000000000000000 --- a/LogitsProcessor_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:951c3c3f41107a837943d8b97b63e30532b612473487d03be2bc04de07e03654 -size 44978 diff --git a/LogitsProcessor_prefill_prefill_H100.png b/LogitsProcessor_prefill_prefill_H100.png deleted file mode 100644 index de89b349b725621db5ff87f1163ef6619f4214e5..0000000000000000000000000000000000000000 --- a/LogitsProcessor_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:b1b94b5b8b0eeb8e1c282e932ac25d5567eca9afc68a168b7788b9ffcad3e292 -size 40673 diff --git a/LogitsProcessor_prefill_prefill_MI300x.png b/LogitsProcessor_prefill_prefill_MI300x.png deleted file mode 100644 index 3c428ba307dfd981da3255d914a6c1798560a41a..0000000000000000000000000000000000000000 --- a/LogitsProcessor_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:6ea1361193bb8021ae5490b0b918a9cde193edc8bb3db37fbe4b68fa5c75d23e -size 41177 diff --git a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_abs_diff.png b/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_abs_diff.png deleted file mode 100644 index c664efdb0c994d085f1fb601f87b6eb37128bd4c..0000000000000000000000000000000000000000 --- a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:3e30c678d6dc3c2273ce0d1d8ea8d126ead569afc79295299109f09a39186072 -size 52375 diff --git a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_decode_1_H100.png b/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_decode_1_H100.png deleted file mode 100644 index 6ebfe7ce50882be0586f82abc464b238f36e8a86..0000000000000000000000000000000000000000 --- a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:e472973f87f4759570114d8c3989c3ef6dd5e920e57d07c1c11b944a4d782cd0 -size 56094 diff --git a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_decode_1_MI300x.png b/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_decode_1_MI300x.png deleted file mode 100644 index 8b5806bdf619b0f29e6523be568b7da83ab213ba..0000000000000000000000000000000000000000 --- a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:7325edea76850c0b00da150b1adf6cda47def20c9f2d13f81032e6d388e903dc -size 51895 diff --git a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_abs_diff.png b/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_abs_diff.png deleted file mode 100644 index 114ae2adb9750fc897652a967e002f339e121212..0000000000000000000000000000000000000000 --- a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a850f7b56ad764a73896b37d163888370f953759e3461717dc262f52c5f31126 -size 51390 diff --git a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_prefill_H100.png b/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_prefill_H100.png deleted file mode 100644 index 25cb2a5056a620ecd988f8377d1a675022535cef..0000000000000000000000000000000000000000 --- a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:b12f509d2cb6613df24ec0602382a0203716f73bc291409f64e349cc6afb8d6a -size 51031 diff --git a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_prefill_MI300x.png b/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_prefill_MI300x.png deleted file mode 100644 index 42de349d4a331c9c7aa248e5da6ae9143b269962..0000000000000000000000000000000000000000 --- a/MergedColumnParallelLinear(weight=bfloat16[28672, 4096])_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:870f1197da82d459fa88eeb714c042ced1b6cf234221c12ea8177da26b9bf225 -size 51089 diff --git a/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_abs_diff.png b/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_abs_diff.png deleted file mode 100644 index 231bcee0ca23461e0a0ef8391902a65a771235b3..0000000000000000000000000000000000000000 --- a/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:2cbecbed1252623f53a81072f159222345ea287d2276526b6606659bbb01b69c -size 50265 diff --git a/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_decode_1_H100.png b/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_decode_1_H100.png deleted file mode 100644 index 7659936ea5fb49c52d33ada182741dbd8881f78b..0000000000000000000000000000000000000000 --- a/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:0157fc0575df5e885da2fbb49776d36939c96f70331be6aa21c2c4bc7e2ce9fc -size 48571 diff --git a/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_decode_1_MI300x.png b/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_decode_1_MI300x.png deleted file mode 100644 index 8fb020870ca0f4cab1caeb8cb5454996df2076e2..0000000000000000000000000000000000000000 --- a/QKVParallelLinear(weight=bfloat16[6144, 4096])_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c6fac0ae1c95864d6750769bb6d409a3ecc44fac3ef46cf5a425e5c1c7ab7a85 -size 50138 diff --git a/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_abs_diff.png b/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_abs_diff.png deleted file mode 100644 index 384a01489a44c446fcd161d0a2a94a3f69f9d356..0000000000000000000000000000000000000000 --- a/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a741bff9286d97e1525e26f731064600bd0b213bb4e4a1789bad9ab593c0bd34 -size 52605 diff --git a/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_prefill_H100.png b/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_prefill_H100.png deleted file mode 100644 index 0e4aa4fe53091263dfc3eb03c1ac237334ca7197..0000000000000000000000000000000000000000 --- a/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:0c685d619a87786e4380bf156a1aded102ee8bc3d61c3cd4f273b943c3d3261b -size 50258 diff --git a/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_prefill_MI300x.png b/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_prefill_MI300x.png deleted file mode 100644 index 4cc897f9b6a4f0b9783593e25dab807f831952cb..0000000000000000000000000000000000000000 --- a/QKVParallelLinear(weight=bfloat16[6144, 4096])_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:44e73334dabf05bea0c6c5f7b5937d2925aa805efbd87618adc4a0d29cc61a6a -size 51242 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_abs_diff.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_abs_diff.png deleted file mode 100644 index e7073ef5aa87a4187dd399aa3b0915a4702abb27..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:de156841c86e2057c0914f8e58d7f0a6065d731d06ee4033ed00ef372c878333 -size 54586 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_decode_1_H100.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_decode_1_H100.png deleted file mode 100644 index 529aadebd2fb0ddad2cabe8801db25b3bff92d45..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:da070eb3a842d4a6febb3ccc6fe92f84afe30f323a00d3be4ba9e1cb07bd15c3 -size 55539 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_decode_1_MI300x.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_decode_1_MI300x.png deleted file mode 100644 index 6d6c0dcc25be9d6c5e4fee3a44f76b537aff5684..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:283bf63822b2c7fe57106850bb80fc37689cc8082790e5496cdb987867976455 -size 51811 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_abs_diff.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_abs_diff.png deleted file mode 100644 index 5d8feb5f81c539cd9cce985c7c7ae766cbf1b8f5..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:3b538c2414ffd196a6f9a8a3c34023c6b5783c5f54ac323ce42c4c7bdab5b046 -size 51867 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_prefill_H100.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_prefill_H100.png deleted file mode 100644 index 3d98b8c77345f295b6eb2345b088f66dee6a4398..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:531f39622fe17a9d201518580d5e549c6dd735e6bdc8cbaadbb37c159aace4f7 -size 49660 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_prefill_MI300x.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_prefill_MI300x.png deleted file mode 100644 index 54da1b92b960b0981e9f1191355cf0cdc0834ee5..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:3db762abc3ae88e50432d8d20bc1bb846d530b33cb2039fa25e251ef4240a680 -size 51269 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_abs_diff.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_abs_diff.png deleted file mode 100644 index 55ba00449c1ca338e8581e1e5bf43ebeb4e4f2a4..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:b4d23e1356bd0dba7cbaafce608850035102d0b4650356aaa5ed6a08fe9b9dc9 -size 56549 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_decode_1_H100.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_decode_1_H100.png deleted file mode 100644 index a75699c94ed350f3f239f862f03f6b9252b496de..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:80dc8c418b8d59f5b9693c98252dbf22b86a20a26b229c19a90f780663d30182 -size 52815 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_decode_1_MI300x.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_decode_1_MI300x.png deleted file mode 100644 index a63a44a9c1d79414c084decdfdd4a6668b17699e..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:2f113af8d0cd96667e26ff18b08880d163c5fe36b4c65df0a241988186d71d07 -size 53658 diff --git a/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_prefill_prefill_MI300x.png b/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_prefill_prefill_MI300x.png deleted file mode 100644 index f9f06dbc059fe5885c76ef4ac85e10661998d950..0000000000000000000000000000000000000000 --- a/RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:39d8cf99387a2ff648a1f83d5f1f8142ed8896d30062c012119000e6c7c569fe -size 50270 diff --git a/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_abs_diff.png b/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_abs_diff.png deleted file mode 100644 index cd219fd9557dd65251b606e821be41d0bf3678ae..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a53d5ed206192df11cebfb78d638ce481d483737847c430c666c34d9b8606220 -size 50915 diff --git a/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_decode_1_H100.png b/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_decode_1_H100.png deleted file mode 100644 index c2ae7007ed9418f4b1eb3689dbd4b8df48320489..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:aee260026d262d23e24c8417780fab4a3bb9acf6f9447c373d90b4328f611e4f -size 50005 diff --git a/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_decode_1_MI300x.png b/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_decode_1_MI300x.png deleted file mode 100644 index 50d5e1b07bc20a38eb97513fd36a540b4a1fa22e..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 14336])_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:284f17ac614cb504fa68e3952880457de4f4303bf0ca6b9ee2623e99f20141a4 -size 49615 diff --git a/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_abs_diff.png b/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_abs_diff.png deleted file mode 100644 index d3cb61d3503a7a93eb63142c6f45832258a0356f..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:5c7d30afc7987d5a3d5d53b5d8c40c70f25552a7dc88c7bc15f26a6de979d09a -size 49356 diff --git a/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_prefill_H100.png b/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_prefill_H100.png deleted file mode 100644 index 0d594e2fecfbc94f758a730078a4f39c06fc64eb..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:e042dcc6a7ef128ab3a591e6dc56188032dfcec0bb4a27d414de479cef81521a -size 48372 diff --git a/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_prefill_MI300x.png b/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_prefill_MI300x.png deleted file mode 100644 index cef1810a4292ba063c8eb85d0d85d31993fd5665..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 14336])_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:99c3576acd1bb9b4ecc61e1ddbd69ef00d245264dc0b615f58dd129597f105eb -size 50102 diff --git a/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_abs_diff.png b/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_abs_diff.png deleted file mode 100644 index 0a3531f58edcc1cb15e4624e4144176e09dd789d..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:8417e1a436602d814a3ecae279e30a3d312bf30572d6456175c2eea46655efcd -size 52083 diff --git a/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_decode_1_H100.png b/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_decode_1_H100.png deleted file mode 100644 index 79739710967a6e10ee89ebbc833745228c91a037..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c458cda0ae57bd697b1267afbea6b4991f942050e714d27e40e148eddfe179fc -size 47190 diff --git a/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_decode_1_MI300x.png b/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_decode_1_MI300x.png deleted file mode 100644 index 081259e317dce1a0cf49aedddded7e2db4eb53cf..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 4096])_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a3b65e31cbf42b38fe35b53d12207f4c08b312b5ef3030c7dbc7af1df8f194d2 -size 49355 diff --git a/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_abs_diff.png b/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_abs_diff.png deleted file mode 100644 index fad48690be2e5f125881a4cec57b29267dc98144..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:0a8ea5efa844aad0b4fefe12a0cb70005700e6153f3de6dae3eb7add0eab91a8 -size 50344 diff --git a/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_prefill_H100.png b/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_prefill_H100.png deleted file mode 100644 index 180fb1fcf503d3d8d9897fbca44d84cf3b8e5e56..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:53e67d6ed25635a5f5b4f2acd82110c3d16cdf3cfcd8f3e2dc61b55e733d8556 -size 51758 diff --git a/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_prefill_MI300x.png b/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_prefill_MI300x.png deleted file mode 100644 index e71a56cf9d44748fb40adef8a9561993d0013730..0000000000000000000000000000000000000000 --- a/RowParallelLinear(weight=bfloat16[4096, 4096])_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c268782d1157dca143160f2dd68626aed6f23dc7af14496ec15318a1b9af14df -size 50223 diff --git a/Sampler_decode_abs_diff.png b/Sampler_decode_abs_diff.png deleted file mode 100644 index e925c8dd767b0140d3357bcaa11dfba1a04e7e71..0000000000000000000000000000000000000000 --- a/Sampler_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:d335e61a2d8071d0a5871612c31bfa33377c99e1d2d4b37447c79f32b5fc0f96 -size 37988 diff --git a/Sampler_decode_decode_1_H100.png b/Sampler_decode_decode_1_H100.png deleted file mode 100644 index 795ecb52aa86c3e4465dccf4e73c8df3b16f5534..0000000000000000000000000000000000000000 --- a/Sampler_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:effcc6ceb0b0b1311c85273e06f40a7cf6804c46e1673d17b8a15615292ddd41 -size 36165 diff --git a/Sampler_decode_decode_1_MI300x.png b/Sampler_decode_decode_1_MI300x.png deleted file mode 100644 index 318649c50a4f083da312976eb3371d0abcd21fa2..0000000000000000000000000000000000000000 --- a/Sampler_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:e3a3a4e486fdb3603b1a9c97fbe7e89144b73840e2156b3d953ea5aa51a66419 -size 35986 diff --git a/Sampler_prefill_abs_diff.png b/Sampler_prefill_abs_diff.png deleted file mode 100644 index 8a487303372691cf1ab9ca93f28a08eed358e69a..0000000000000000000000000000000000000000 --- a/Sampler_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:59f27efe07dfa1991af4105130660f56d629b82980a43e13472f1bd2a850581d -size 42877 diff --git a/Sampler_prefill_prefill_H100.png b/Sampler_prefill_prefill_H100.png deleted file mode 100644 index 232231936cc273b606d9d208150a6a70940ac2ca..0000000000000000000000000000000000000000 --- a/Sampler_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:acd905c980d91f41e91bb5705187574ed20bd726cc4cbcf6d7a342cef165c0b7 -size 37611 diff --git a/Sampler_prefill_prefill_MI300x.png b/Sampler_prefill_prefill_MI300x.png deleted file mode 100644 index ca923fad4fc175dbad52173e0c7093b373c5cb4f..0000000000000000000000000000000000000000 --- a/Sampler_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:7dabfdeda2ee8f79793c3b472f5ac945b3729afcc3e585b3ac2e6ac25ec05a63 -size 37650 diff --git a/SiluAndMul_decode_abs_diff.png b/SiluAndMul_decode_abs_diff.png deleted file mode 100644 index 9c02f06565b46b09895a7a91d0e0024eb7733312..0000000000000000000000000000000000000000 --- a/SiluAndMul_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f421e922440c6f85717f23fda6416f27f607568df60cdb9c66fb4b40d634b5b5 -size 41017 diff --git a/SiluAndMul_decode_decode_1_H100.png b/SiluAndMul_decode_decode_1_H100.png deleted file mode 100644 index fb3311d380a1977edea68dbe6dd1c2a87680381b..0000000000000000000000000000000000000000 --- a/SiluAndMul_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:6644d4f519ff2d8aff5fb59001fd2eecf25d33fc5d7bb32b42eccbbdef24e6ab -size 39337 diff --git a/SiluAndMul_decode_decode_1_MI300x.png b/SiluAndMul_decode_decode_1_MI300x.png deleted file mode 100644 index 40623bcd32ff8a0891e9dfaa06137614f95ff948..0000000000000000000000000000000000000000 --- a/SiluAndMul_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:d49179082de21a978caf09b9bc29eb9a0d05c09a0fcaec61353e4bea67dfe4ca -size 38315 diff --git a/SiluAndMul_prefill_abs_diff.png b/SiluAndMul_prefill_abs_diff.png deleted file mode 100644 index 11f1585c43a52c8381822c4c438c00a52a20bce6..0000000000000000000000000000000000000000 --- a/SiluAndMul_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:6eb4be6df557ff103765e5f144961065b3e6e9a8998a7d8f8dae93d77aafe001 -size 40436 diff --git a/SiluAndMul_prefill_prefill_H100.png b/SiluAndMul_prefill_prefill_H100.png deleted file mode 100644 index 0ee1e0dc6bd1bb7132cb4d9d138697c1870bebaa..0000000000000000000000000000000000000000 --- a/SiluAndMul_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:2ea6b074cd37defbad338fef5b56e8f4e0b0abf771cb074d50b640099e274140 -size 38282 diff --git a/SiluAndMul_prefill_prefill_MI300x.png b/SiluAndMul_prefill_prefill_MI300x.png deleted file mode 100644 index 03cfee198a9b413531df85787b2c5ac42e2118b2..0000000000000000000000000000000000000000 --- a/SiluAndMul_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bd6018658af7bb7c95fe536134f72eaa7e85ef6f621bbaf8652c37cebc5f3bf5 -size 38734 diff --git a/decode_h100.png b/decode_h100.png deleted file mode 100644 index 235292a9e898c9cd8f570d63d6801478054aef42..0000000000000000000000000000000000000000 --- a/decode_h100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:69be199055247a195d06cc30af913a0013c49cbc496d24538462bc59fd60a125 -size 26754 diff --git a/decode_mi300x.png b/decode_mi300x.png deleted file mode 100644 index 716e0bd737537a297e05d7e1af3951e28eddcd59..0000000000000000000000000000000000000000 --- a/decode_mi300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:ae8513f2257f74a2e15f990440d092a605ff11d42312ca754ac4770c34f0fc6a -size 27188 diff --git a/decode_speedup.png b/decode_speedup.png deleted file mode 100644 index 2a51b7d5cc2b0cc9f92583b080b842a3c38e3908..0000000000000000000000000000000000000000 --- a/decode_speedup.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:561ec5e7771ea201cc48f0a77f11830e6edbcfab743c1b1eab03f707035ec968 -size 31457 diff --git a/others_decode_abs_diff.png b/others_decode_abs_diff.png deleted file mode 100644 index 00d728181a8f8ebc86d198f1e272f104ae51be5f..0000000000000000000000000000000000000000 --- a/others_decode_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:338d8dfa13d42de53d3625a2fd5fb798eb9bab3c5ce749adf57b3e499e94503e -size 43071 diff --git a/others_decode_decode_1_H100.png b/others_decode_decode_1_H100.png deleted file mode 100644 index 26cc19a8ff97d16c9f5a6b12040cb4dd3c69b02c..0000000000000000000000000000000000000000 --- a/others_decode_decode_1_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:16e2efa7ae870ff9c09abf91f39eccf53a7152946cb518fd77fbf6db6d97a6c3 -size 38327 diff --git a/others_decode_decode_1_MI300x.png b/others_decode_decode_1_MI300x.png deleted file mode 100644 index 8ccbb5c8833a63f9caa49217af947b90c7e9e268..0000000000000000000000000000000000000000 --- a/others_decode_decode_1_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:80b88efd2c78be9e8e1c211113a3c165e2410c0cba1c964a92579046de35420e -size 40559 diff --git a/others_prefill_abs_diff.png b/others_prefill_abs_diff.png deleted file mode 100644 index ae5e4607419a772543eb742c3fc447420fe44baa..0000000000000000000000000000000000000000 --- a/others_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:dbefb21bbb5b91db0e14e4365ceafbcf507c41d3a924739334b03d3a377de7ec -size 41740 diff --git a/others_prefill_prefill_H100.png b/others_prefill_prefill_H100.png deleted file mode 100644 index 4856d537d51daa058400d7c4c5b09f9b1c162bcd..0000000000000000000000000000000000000000 --- a/others_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:8944c869e6d0d75c6b578e5e17ea64d79e4870453df04c2f0eb3626d3c8a94bc -size 39663 diff --git a/others_prefill_prefill_MI300x.png b/others_prefill_prefill_MI300x.png deleted file mode 100644 index 8f9fdf3eecd4a36b1df4001994707e158b224399..0000000000000000000000000000000000000000 --- a/others_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:cbae957457034bfe040221d962f83f053800f83ee1a8e309f278d08ae6dd81db -size 43340 diff --git a/prefill_h100.png b/prefill_h100.png deleted file mode 100644 index 72f2935d5323dcedd855bbb5f5559a3cd60a324b..0000000000000000000000000000000000000000 --- a/prefill_h100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:71945c3ab7e54b6ef727b648439ef727014583ee933e201c5abc0bb59efe6da4 -size 26227 diff --git a/prefill_mi300x.png b/prefill_mi300x.png deleted file mode 100644 index 422dfbd0c6867fc871da134d6cce92f0c66f11f6..0000000000000000000000000000000000000000 --- a/prefill_mi300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:44ec1f416b72a6228e5d9ad538d147768c85401e9815729035933d7e0623aa23 -size 27334 diff --git a/prefill_speedup.png b/prefill_speedup.png deleted file mode 100644 index 350f2b58766c96e51bdb0f2d7abb1a037d39937d..0000000000000000000000000000000000000000 --- a/prefill_speedup.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:4b71dce95ba2411d2572848109984a55aefc990406726bfeac5302fba16e8a22 -size 27945 diff --git a/profiling_bs1_pl1024/H100decode.pkl b/profiling_bs1_pl1024/H100decode.pkl deleted file mode 100644 index 1f99a76a1e48c88658220a8362f520f294252808..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/H100decode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:18a56ec3543b1348662da7f0ee74c57bf9cd76d9e4e38a32192892654479ac4d -size 1133 diff --git a/profiling_bs1_pl1024/H100decode_steps.png b/profiling_bs1_pl1024/H100decode_steps.png deleted file mode 100644 index fbe467a5db71aae7fdd304a1c1150f529fb7843a..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/H100decode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:fbcbe5727b60006fb5703a849fd4bee673738ed1afd9443688a73b7e0dfd090d -size 71822 diff --git a/profiling_bs1_pl1024/H100prefill.pkl b/profiling_bs1_pl1024/H100prefill.pkl deleted file mode 100644 index bfdaae104232f5a4be704e5f85311a7f28d2ffec..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/H100prefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f566d14bf6a8e96883145c9f751c1881276d1f3fb58c860b13d123744b2bdbe8 -size 1149 diff --git a/profiling_bs1_pl1024/H100prefill.png b/profiling_bs1_pl1024/H100prefill.png deleted file mode 100644 index 6279b2129194ca72612eafc5ec54bbc29076b241..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/H100prefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:b760196ea1c482825e68a99fbaacb09b687e2b0aa7fbc8867cdf478442a80e3c -size 65476 diff --git a/profiling_bs1_pl1024/MI300xdecode.pkl b/profiling_bs1_pl1024/MI300xdecode.pkl deleted file mode 100644 index 48a7c5340a54c18a6cb39c9d780c1781cbab482e..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/MI300xdecode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:2b3ba217ada646eb1d8c67824f613abff09ce245ec3ecb27948597937634e7cd -size 1133 diff --git a/profiling_bs1_pl1024/MI300xdecode_steps.png b/profiling_bs1_pl1024/MI300xdecode_steps.png deleted file mode 100644 index 8726f354e90e21883e5576a7bacff408df7988ae..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/MI300xdecode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c5281ce5518c3f29246fe1f47913306683f8fc2ad383aadef44b2a39117793e7 -size 71791 diff --git a/profiling_bs1_pl1024/MI300xprefill.pkl b/profiling_bs1_pl1024/MI300xprefill.pkl deleted file mode 100644 index d6375c1c699b07dfa9b6d9cb86927ebfcaa815df..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/MI300xprefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:52d330a6d72d15f43c82ee2ecde0b08549fa0521a738f654c760eabba7ddff4f -size 1149 diff --git a/profiling_bs1_pl1024/MI300xprefill.png b/profiling_bs1_pl1024/MI300xprefill.png deleted file mode 100644 index 1bb5f160c0cd5e2aa4e69bf73ce1ae1463562422..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/MI300xprefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:2b2a93f56de690728a98db2bf0e641e9a79cba2c97a43e086600743f377cdcc1 -size 65481 diff --git a/profiling_bs1_pl1024/decode_comparison.csv b/profiling_bs1_pl1024/decode_comparison.csv deleted file mode 100644 index c5d02911962e35695e87703d89514ae80325a919..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/decode_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,decode_1_H100,decode_1_MI300x,abs_diff -Attention,8.752717489220126,7.27695149430903,-1.4757659949110966 -Llama3RotaryEmbedding,1.8000202471177884,2.104369711457971,0.3043494643401825 -LogitsProcessor,4.964075664737471,4.514901920960399,-0.449173743777072 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",37.824289906697395,35.46849998215996,-2.355789924537433 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.576314761464358,10.226939006282445,0.6506242448180863 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,2.9357176888315397,5.175429327375672,2.239711638544132 -RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,0.04621376037593292,0.07626616238825962,0.030052402012326694 -"RowParallelLinear(weight=bfloat16[4096, 14336])",19.938635218431514,19.464116135902668,-0.4745190825288468 -"RowParallelLinear(weight=bfloat16[4096, 4096])",8.381148560652987,7.621727382262613,-0.7594211783903742 -Sampler,1.731400248341767,3.2102290660877353,1.4788288177459683 -SiluAndMul,4.0082858755763136,4.787734424960685,0.7794485493843712 -others,0.041180578552811514,0.07283538585257543,0.031654807299763915 diff --git a/profiling_bs1_pl1024/decode_comparison.pkl b/profiling_bs1_pl1024/decode_comparison.pkl deleted file mode 100644 index d4325f38e15247a7108f1e48af354536a6fa8bbe..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/decode_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:ede17193550fd6395e6be92cb4cf0da723d514f9ddb731be0a76a3977c5ebde3 -size 1520 diff --git a/profiling_bs1_pl1024/prefill_comparison.csv b/profiling_bs1_pl1024/prefill_comparison.csv deleted file mode 100644 index 5780279d6dc3c7c830e86c50d58e412698cfcf89..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/prefill_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,prefill_H100,prefill_MI300x,abs_diff -Attention,5.315293851953808,16.011133629401503,10.695839777447695 -Llama3RotaryEmbedding,1.796983334944434,3.606864748181656,1.809881413237222 -LogitsProcessor,1.5295391766961255,0.5252318818660545,-1.0043072948300709 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",43.41207624128187,34.531058283263,-8.881017958018873 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",11.40034029870664,9.22755468782225,-2.1727856108843895 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,2.841601131933758,3.3789828271301117,0.5373816951963537 -"RowParallelLinear(weight=bfloat16[4096, 14336])",20.329310428897845,21.313634076253603,0.9843236473557582 -"RowParallelLinear(weight=bfloat16[4096, 4096])",6.488699910092019,7.525435921409029,1.0367360113170099 -Sampler,0.5329721045738438,0.5531418514801119,0.02016974690626805 -SiluAndMul,6.179633746887224,3.1609146031265904,-3.0187191437606335 -others,0.045026702926414884,0.05639850957638423,0.011371806649969347 -vocab_embed_ops,0.128523071106005,0.10964898048974996,-0.018874090616255043 diff --git a/profiling_bs1_pl1024/prefill_comparison.pkl b/profiling_bs1_pl1024/prefill_comparison.pkl deleted file mode 100644 index a1dfc3e1ea6d09c43636e328806372320114131e..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1024/prefill_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:0162910352fc1aa9d5f9093de307927126b7c195b4f9ead5cd13779ec7aee02d -size 1483 diff --git a/profiling_bs1_pl128/H100decode.pkl b/profiling_bs1_pl128/H100decode.pkl deleted file mode 100644 index e2fad026fbc92bfbce8a742f63132adf004f5e8c..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/H100decode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:232a9917182ebd5eab8e0112687e0708a2db5b71cbbcbef9f878b5975621e5f0 -size 1133 diff --git a/profiling_bs1_pl128/H100decode_steps.png b/profiling_bs1_pl128/H100decode_steps.png deleted file mode 100644 index 260375d46c89b47e21a0f3bfe534ff4a2bf00f93..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/H100decode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:9ec8e9b14ce546935c7c750470fced0541779dc578f03672480829e7ded48517 -size 71958 diff --git a/profiling_bs1_pl128/H100prefill.pkl b/profiling_bs1_pl128/H100prefill.pkl deleted file mode 100644 index bd3995886c0bd9d403009571c95ffa66ed08eaf4..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/H100prefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:4b05faa251a6026f0bcb7213ea680318ea294d5e77d94c7e619dbee792cffd40 -size 1149 diff --git a/profiling_bs1_pl128/H100prefill.png b/profiling_bs1_pl128/H100prefill.png deleted file mode 100644 index f36226769d28d493307124db30dc862fbe6e94c7..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/H100prefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:ef8ac45fe3046cb920cfcd777c1a917f56091095cdcacce69e1c05a5e8b9e3ec -size 65451 diff --git a/profiling_bs1_pl128/MI300xdecode.pkl b/profiling_bs1_pl128/MI300xdecode.pkl deleted file mode 100644 index e8a50992e3f86cc77623770b5bb6ca8298374038..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/MI300xdecode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c3458dd3380674c3169e1142e5bb3a15ed596c76674522a22c63d067c9df073a -size 1133 diff --git a/profiling_bs1_pl128/MI300xdecode_steps.png b/profiling_bs1_pl128/MI300xdecode_steps.png deleted file mode 100644 index 42fab1633f88f8554ce45d9c3b0951e0b802b83f..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/MI300xdecode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a0d0a83a631e817a49ccb0b75ea6797469ef12095cad52cbf100df167dea1ea0 -size 71696 diff --git a/profiling_bs1_pl128/MI300xprefill.pkl b/profiling_bs1_pl128/MI300xprefill.pkl deleted file mode 100644 index 37df328197f6433d9df464a98f47f073c27841d4..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/MI300xprefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:ce7439d2f829b64beafc7cd258e994b91866b30c6d42c3502aa805f5b6a67890 -size 1132 diff --git a/profiling_bs1_pl128/MI300xprefill.png b/profiling_bs1_pl128/MI300xprefill.png deleted file mode 100644 index d98a99ec04b39b852688c773556100a234842e0d..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/MI300xprefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:7beb049e3736d9b1fd2ec1375ed7fe2748005ba1ea9a6a5368638724cfc3b9ae -size 69214 diff --git a/profiling_bs1_pl128/decode_comparison.csv b/profiling_bs1_pl128/decode_comparison.csv deleted file mode 100644 index a74b699f91f9a0486a80cfc91284e3d51535c9ea..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/decode_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,decode_1_H100,decode_1_MI300x,abs_diff -Attention,6.854378547570919,7.623201194795887,0.768822647224968 -Llama3RotaryEmbedding,1.81774562552658,2.1432639014771704,0.32551827595059035 -LogitsProcessor,5.075146841646092,4.58921266236861,-0.48593417927748206 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",38.652132986747134,35.20547598606534,-3.446657000681796 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.785950245233131,9.46617176958518,-0.3197784756479507 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,3.0304211602626046,5.20425249178647,2.173831331523865 -RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,0.0468820505350108,0.07704894566333666,0.030166895128325856 -"RowParallelLinear(weight=bfloat16[4096, 14336])",20.407072199090585,19.958467048318322,-0.4486051507722628 -"RowParallelLinear(weight=bfloat16[4096, 4096])",8.50892363706815,7.982606971699157,-0.5263166653689932 -Sampler,1.7628618245720273,3.143947047999449,1.3810852234274218 -SiluAndMul,4.016277846567755,4.543894850198202,0.5276170036304473 -others,0.042207035180003474,0.062457130042884695,0.02025009486288122 diff --git a/profiling_bs1_pl128/decode_comparison.pkl b/profiling_bs1_pl128/decode_comparison.pkl deleted file mode 100644 index 30798b70bda39230c3bc71fcfd25b3fa541459ea..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/decode_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:8a5b0a41877b5248c5f8183b69bfc5b52097cea7162334a8406175bae1ff27df -size 1520 diff --git a/profiling_bs1_pl128/prefill_comparison.csv b/profiling_bs1_pl128/prefill_comparison.csv deleted file mode 100644 index b32d3a027cf7af42e3d7e7cf6aeb91ddd7c144e3..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/prefill_comparison.csv +++ /dev/null @@ -1,14 +0,0 @@ -name,prefill_H100,prefill_MI300x,abs_diff -Attention,5.848365577953972,17.819456162737453,11.97109058478348 -Llama3RotaryEmbedding,1.7251192777954603,4.834478626995694,3.109359349200234 -LogitsProcessor,4.540207923847165,1.9085394978140058,-2.631668426033159 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",37.7209583312918,30.437784289662257,-7.28317404162954 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.895869936136904,9.277510581328052,-0.6183593548088524 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,2.9142125957824905,2.927573395252407,0.013360799469916529 -RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,,0.044807943956731024, -"RowParallelLinear(weight=bfloat16[4096, 14336])",23.992343837415707,18.76974919753904,-5.222594639876668 -"RowParallelLinear(weight=bfloat16[4096, 4096])",8.607495036906135,9.569063648344288,0.9615686114381532 -Sampler,1.565819497993548,1.5190539444355344,-0.046765553558013595 -SiluAndMul,3.062675209068646,2.848634712334306,-0.2140404967343401 -others,0.04454403836961201,0.0433479996002222,-0.0011960387693898053 -vocab_embed_ops,0.08238873743857462,, diff --git a/profiling_bs1_pl128/prefill_comparison.pkl b/profiling_bs1_pl128/prefill_comparison.pkl deleted file mode 100644 index d98d61c9801af1a27fcf8a24f75ca8b08823f666..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl128/prefill_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:b17c35c0b584d766cbda4dfcb6c16d2964edf0b5629d9b53c4d2987e8f0dadea -size 1526 diff --git a/profiling_bs1_pl1536/H100decode.pkl b/profiling_bs1_pl1536/H100decode.pkl deleted file mode 100644 index ec950ddeba9c702d3e2da40233548842fb682e25..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/H100decode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a8e88b351d543763c88fb1c5c1d79d04290a882df4b04cb4c0ab89ebf725c809 -size 1133 diff --git a/profiling_bs1_pl1536/H100decode_steps.png b/profiling_bs1_pl1536/H100decode_steps.png deleted file mode 100644 index bf893dcf1b42cec3b4a083959177065e5d48ff57..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/H100decode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bd3b1991d308d7ca30ee085f4507a0a9f39351b02710b9b878921cb7dc08d8b7 -size 71850 diff --git a/profiling_bs1_pl1536/H100prefill.pkl b/profiling_bs1_pl1536/H100prefill.pkl deleted file mode 100644 index 451c2ee76337757fa32d69243358d45a517e9a2a..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/H100prefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bb7876b2f7c4e20f362b6c1b1b5892971f7dbc38b284d3c25b8123d97f125fe5 -size 1149 diff --git a/profiling_bs1_pl1536/H100prefill.png b/profiling_bs1_pl1536/H100prefill.png deleted file mode 100644 index 7e887a0f2e52f699155ce5047c31ac170467c6cc..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/H100prefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f9534069fab46b5c0298064547cf9ef3e7c702d6db978c92aa8bed3b29ecc11f -size 65558 diff --git a/profiling_bs1_pl1536/MI300xdecode.pkl b/profiling_bs1_pl1536/MI300xdecode.pkl deleted file mode 100644 index 8aba1d5c870eeb08b73f43485ebf2f8af9317d70..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/MI300xdecode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:22b1ba54fee87c9612e916aa2808a1f691121cd333ab37d306303c791382d474 -size 1133 diff --git a/profiling_bs1_pl1536/MI300xdecode_steps.png b/profiling_bs1_pl1536/MI300xdecode_steps.png deleted file mode 100644 index cef28d9fe1d93e648e861fd590cba68c3206e532..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/MI300xdecode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:3c1837cc7b38a38b53d9591231aa74a0d4ed1c27040851e6e3772e696ede1f1e -size 71646 diff --git a/profiling_bs1_pl1536/MI300xprefill.pkl b/profiling_bs1_pl1536/MI300xprefill.pkl deleted file mode 100644 index 95a0393ad5978f62bd49a775ff715ccdbadb2f34..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/MI300xprefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:bb8240282286ef7fb58bafb017dc2f1856190f43796205b58a3ed40437a61856 -size 1149 diff --git a/profiling_bs1_pl1536/MI300xprefill.png b/profiling_bs1_pl1536/MI300xprefill.png deleted file mode 100644 index cf771bf5aed61b8cf21aa98065fbc6aa68ad50bd..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/MI300xprefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:d8ebf13d7792d0355d8eae5f4cf99e2ba2e4cbeba0504c985d7176555ea1fa5d -size 65520 diff --git a/profiling_bs1_pl1536/decode_comparison.csv b/profiling_bs1_pl1536/decode_comparison.csv deleted file mode 100644 index 1c491fa910bfa124a1e6ec1a04166a44ab24347d..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/decode_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,decode_1_H100,decode_1_MI300x,abs_diff -Attention,9.483617729518452,7.598486548526112,-1.8851311809923397 -Llama3RotaryEmbedding,1.6607296965324518,2.131330682616359,0.4706009860839073 -LogitsProcessor,4.941050766659664,4.599842707760885,-0.34120805889877914 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",37.43929849955707,35.85625615291251,-1.583042346644561 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.482533264536395,9.674445360747388,0.19191209621099325 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,3.078096889526124,5.296593831003937,2.2184969414778135 -RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,0.048401384462242354,0.07861138187206942,0.03020999740982707 -"RowParallelLinear(weight=bfloat16[4096, 14336])",19.757810430954965,18.882085706206716,-0.8757247247482489 -"RowParallelLinear(weight=bfloat16[4096, 4096])",8.381144921050947,7.9135920920354135,-0.4675528290155331 -Sampler,1.7479292426553181,3.226106204220072,1.478176961564754 -SiluAndMul,3.938748276271449,4.667511718758141,0.728763442486692 -others,0.040638898274901594,0.07513761334038274,0.03449871506548115 diff --git a/profiling_bs1_pl1536/decode_comparison.pkl b/profiling_bs1_pl1536/decode_comparison.pkl deleted file mode 100644 index fe980642ac3779a9eb4797ce8c3f6249c9172fe4..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/decode_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:984e0cbd6a2c5ae6becdb657e852f7170b687ca915e72cb38c98a9a8093c44df -size 1520 diff --git a/profiling_bs1_pl1536/prefill_comparison.csv b/profiling_bs1_pl1536/prefill_comparison.csv deleted file mode 100644 index 3fb5514e498c45e89616cfc5b6cead9486a69f64..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/prefill_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,prefill_H100,prefill_MI300x,abs_diff -Attention,5.725389884282574,21.13655746796299,15.411167583680417 -Llama3RotaryEmbedding,1.7857420562193158,3.1890826435170294,1.4033405872977136 -LogitsProcessor,1.017143181316883,0.35926109541711065,-0.6578820858997723 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",43.29349359545625,35.02980749731597,-8.263686098140276 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.751519197825727,8.501444341490565,-1.2500748563351625 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,2.87586431034889,2.772191898445982,-0.10367241190290821 -"RowParallelLinear(weight=bfloat16[4096, 14336])",21.449999372013618,19.56647746212289,-1.883521909890728 -"RowParallelLinear(weight=bfloat16[4096, 4096])",7.4886429722468115,5.775504701351854,-1.7131382708949578 -Sampler,0.36281806980696546,0.3658306067798142,0.003012536972848756 -SiluAndMul,6.082842551397157,3.1813915450811434,-2.9014510063160133 -others,0.042990959395247705,0.04396077893029697,0.0009698195350492655 -vocab_embed_ops,0.12355384969054385,0.07848996158435963,-0.045063888106184224 diff --git a/profiling_bs1_pl1536/prefill_comparison.pkl b/profiling_bs1_pl1536/prefill_comparison.pkl deleted file mode 100644 index 5a73df6b271cee0a3492b69fc2a32b4e5cfe8c23..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl1536/prefill_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:e46fcbcf6fc0e9e991e097da294fa03f7621a0ff26139cfcd7346be2819171c3 -size 1483 diff --git a/profiling_bs1_pl2048/H100decode.pkl b/profiling_bs1_pl2048/H100decode.pkl deleted file mode 100644 index 6b25258074b240475efbef07c7b7ffd4e44064c4..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/H100decode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:46d0ef803d93ebf1f08c95f76d1ca185458027fbb8be27a29b69dc6babd68a76 -size 1133 diff --git a/profiling_bs1_pl2048/H100decode_steps.png b/profiling_bs1_pl2048/H100decode_steps.png deleted file mode 100644 index bbb0b938642274643117cbf823f8eca89337b17d..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/H100decode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c6bfab1952e9091991395141536b38eda5ceeb1eb2e6e89d5dc7e432f4f1fce5 -size 71869 diff --git a/profiling_bs1_pl2048/H100prefill.pkl b/profiling_bs1_pl2048/H100prefill.pkl deleted file mode 100644 index fd4d5e61cda4d75fbe9b77499f61b998b3ac3f61..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/H100prefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:890c5c68d401ce1d2a64ac2fb5d1359b8cf4c9d8cfa2980b6aee74e6250e35cd -size 1149 diff --git a/profiling_bs1_pl2048/H100prefill.png b/profiling_bs1_pl2048/H100prefill.png deleted file mode 100644 index a9019880e338e4c658260204a113761f004120df..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/H100prefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:9dd124fa321666b01f84f3af59008a0425b759229fd05c1abc76f42b3098d878 -size 65444 diff --git a/profiling_bs1_pl2048/MI300xdecode.pkl b/profiling_bs1_pl2048/MI300xdecode.pkl deleted file mode 100644 index 8c2bbd4ea50ec2b4c302c3750fc07f6edee677c8..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/MI300xdecode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:0080f9fd86520496085195746ebff186382f1555c381b77f3344024eeddd7153 -size 1133 diff --git a/profiling_bs1_pl2048/MI300xdecode_steps.png b/profiling_bs1_pl2048/MI300xdecode_steps.png deleted file mode 100644 index 2657b8a1110525105364eaf654b14ece0e6f8795..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/MI300xdecode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:47ca1036d98173b206c395b269e2c0efb09cd40ca7b0fc5819b763c20e7b2a4d -size 71740 diff --git a/profiling_bs1_pl2048/MI300xprefill.pkl b/profiling_bs1_pl2048/MI300xprefill.pkl deleted file mode 100644 index 92d27ab0cf1033f440606dc4d0a18effa14d216b..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/MI300xprefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:617a7ef44243bf3b885e89fb6373fc31c3fc8ef8a28f228dc785eb840c17a1f2 -size 1149 diff --git a/profiling_bs1_pl2048/MI300xprefill.png b/profiling_bs1_pl2048/MI300xprefill.png deleted file mode 100644 index 4db5cbf14b13f503242a72c1485fb3e384bc3b9e..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/MI300xprefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:9e631a1c7029008b75f4e929d693e247e3c5630505bd5c36def37e0b99db85d9 -size 65565 diff --git a/profiling_bs1_pl2048/decode_comparison.csv b/profiling_bs1_pl2048/decode_comparison.csv deleted file mode 100644 index 73fc592593928a94e722181ae03dfaed8734573b..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/decode_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,decode_1_H100,decode_1_MI300x,abs_diff -Attention,10.431836463519645,7.674151678857323,-2.757684784662321 -Llama3RotaryEmbedding,1.7333178765562836,2.1757345407444877,0.44241666418820413 -LogitsProcessor,4.931779492636019,4.584755769052253,-0.34702372358376543 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",36.53517257642745,35.7003415580844,-0.8348310183430456 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.450435047807359,9.746143482157347,0.2957084343499883 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,2.9514035528290163,5.235026143898585,2.2836225910695687 -RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,0.04489645754478734,0.07783741765778085,0.03294096011299351 -"RowParallelLinear(weight=bfloat16[4096, 14336])",19.83281171434743,19.1229496321115,-0.7098620822359294 -"RowParallelLinear(weight=bfloat16[4096, 4096])",8.319421289197242,7.666690669139791,-0.6527306200574516 -Sampler,1.7282585219655353,3.2277195597310224,1.4994610377654871 -SiluAndMul,3.9984915470514317,4.719852098495739,0.7213605514443073 -others,0.04217546011783053,0.06879745006979515,0.026621989951964624 diff --git a/profiling_bs1_pl2048/decode_comparison.pkl b/profiling_bs1_pl2048/decode_comparison.pkl deleted file mode 100644 index bbf4ce95ff08a786a9615cf354dfdbc794e03196..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/decode_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:1eb06d125f6df32e3123427a40d1826d58c974a383b72e12649abd0b72d4c67b -size 1520 diff --git a/profiling_bs1_pl2048/prefill_comparison.csv b/profiling_bs1_pl2048/prefill_comparison.csv deleted file mode 100644 index 7dcd18429e4c98e3c2dd645a5f35675751153279..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/prefill_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,prefill_H100,prefill_MI300x,abs_diff -Attention,6.595171438841731,26.09203557691796,19.49686413807623 -Llama3RotaryEmbedding,1.8016537430807844,2.766166385861084,0.9645126427802997 -LogitsProcessor,0.7718868383565284,0.26810542465192677,-0.5037814137046016 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",43.06272779789141,32.733081828491585,-10.329645969399827 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",10.206377685608299,7.440274157260589,-2.76610352834771 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,2.877083603112283,2.3348517675593596,-0.5422318355529234 -"RowParallelLinear(weight=bfloat16[4096, 14336])",20.398294463679235,19.44528567543132,-0.9530087882479137 -"RowParallelLinear(weight=bfloat16[4096, 4096])",7.653270666665546,5.753078995573151,-1.9001916710923945 -Sampler,0.27369702137910334,0.2781171389368573,0.004420117557753955 -SiluAndMul,6.188128309643855,2.789111036162593,-3.3990172734812623 -others,0.045881045744944365,0.03594206521152666,-0.009938980533417706 -vocab_embed_ops,0.12582738599630072,0.06394994794203644,-0.06187743805426428 diff --git a/profiling_bs1_pl2048/prefill_comparison.pkl b/profiling_bs1_pl2048/prefill_comparison.pkl deleted file mode 100644 index 404f32be9088483a0de75969b7f2ebe753c63065..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl2048/prefill_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:54cf18d9ec1d52c6d341d3fd74eecc286790b575bde3ad6e4ff84ab361056b3e -size 1483 diff --git a/profiling_bs1_pl256/H100decode.pkl b/profiling_bs1_pl256/H100decode.pkl deleted file mode 100644 index 3c4534c1e66e9335837581bcd49c30a9da621c39..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/H100decode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:274d69894b3ee4588c5294b9ce495ddca1102607dff797b9db9bf736c1fd1c94 -size 1133 diff --git a/profiling_bs1_pl256/H100decode_steps.png b/profiling_bs1_pl256/H100decode_steps.png deleted file mode 100644 index 720222efeddead0db9d98e164f7548cd9b21355c..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/H100decode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:cd97942f63a62e6522b82bf12a59f804f95a7acf6f435139f74d1daf25401f3c -size 71939 diff --git a/profiling_bs1_pl256/H100prefill.pkl b/profiling_bs1_pl256/H100prefill.pkl deleted file mode 100644 index e4fff378b78beeebe2f2155283cc536fc1f000dd..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/H100prefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f38964142b5c09c08f8154c3b0876b2130b4a1e30ac8d5b7ab47df83940b48b6 -size 1149 diff --git a/profiling_bs1_pl256/H100prefill.png b/profiling_bs1_pl256/H100prefill.png deleted file mode 100644 index e29ebd0498e0ddd9007a8b826f11c778843f8754..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/H100prefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:30e25e24e7c6f02d07d0154d68e76726cc14000c419c23816023cfab72839f88 -size 65212 diff --git a/profiling_bs1_pl256/MI300xdecode.pkl b/profiling_bs1_pl256/MI300xdecode.pkl deleted file mode 100644 index 49c577a131c2f4c0c01088ff40a16093f70180a1..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/MI300xdecode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:7117edb56338839eb1e449499eb6ed44fedcbfb5901587ffcfaa3c6aa8610e33 -size 1133 diff --git a/profiling_bs1_pl256/MI300xdecode_steps.png b/profiling_bs1_pl256/MI300xdecode_steps.png deleted file mode 100644 index b0331ec3d82cb3fb89494b8783b6d1dfff20bcb8..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/MI300xdecode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:d0cb50db6034ae71a110eb10e67ea6953d640d5ae0fdb388b54902b7ccc54b9d -size 71714 diff --git a/profiling_bs1_pl256/MI300xprefill.pkl b/profiling_bs1_pl256/MI300xprefill.pkl deleted file mode 100644 index eec86d975171bd6084906fac9d26bc33b555b869..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/MI300xprefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:191a6d5de3265e90f2f8b694255538a27d8cfcf7f7f5f5dca06a4e38d91449fc -size 1132 diff --git a/profiling_bs1_pl256/MI300xprefill.png b/profiling_bs1_pl256/MI300xprefill.png deleted file mode 100644 index f10fb02e333434e7d00efd79d4a33f07a7c93e41..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/MI300xprefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:28e73e4e91be3f614ee6eba446446984f24e14a933cde121640cbbb1c62eda5c -size 69243 diff --git a/profiling_bs1_pl256/decode_comparison.csv b/profiling_bs1_pl256/decode_comparison.csv deleted file mode 100644 index ecaf1f5cb1b514d80b0e507131147c7929c08d3d..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/decode_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,decode_1_H100,decode_1_MI300x,abs_diff -Attention,7.66453436872326,7.376082311989508,-0.2884520567337514 -Llama3RotaryEmbedding,1.7494507609966288,2.1897836926723557,0.44033293167572696 -LogitsProcessor,5.112673045317336,4.554674810701033,-0.5579982346163028 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",37.83081076377858,35.74775400769498,-2.0830567560836 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.760101441762473,9.836106419316247,0.07600497755377411 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,3.0871978635252053,5.302130874418529,2.2149330108933234 -RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,0.04648338109080121,0.07537459921307942,0.028891218122278205 -"RowParallelLinear(weight=bfloat16[4096, 14336])",20.27652623002319,19.101571715463848,-1.1749545145593423 -"RowParallelLinear(weight=bfloat16[4096, 4096])",8.522408056297714,7.76680708946369,-0.7556009668340238 -Sampler,1.8217729154778657,3.2348875322467734,1.4131146167689077 -SiluAndMul,4.084844495629638,4.747850534574219,0.6630060389445811 -others,0.043196677377310214,0.0669764122457414,0.023779734868431185 diff --git a/profiling_bs1_pl256/decode_comparison.pkl b/profiling_bs1_pl256/decode_comparison.pkl deleted file mode 100644 index 6d447986ceedd34f17c617ccad935fe203467b1b..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/decode_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:04d12d513710d554281a00642c2f40ce901c8a24d98747737c868bb8dc08b275 -size 1520 diff --git a/profiling_bs1_pl256/prefill_comparison.csv b/profiling_bs1_pl256/prefill_comparison.csv deleted file mode 100644 index c94f37c54dd4e4d8cc9397bed943981d5910d252..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/prefill_comparison.csv +++ /dev/null @@ -1,14 +0,0 @@ -name,prefill_H100,prefill_MI300x,abs_diff -Attention,5.598502664353409,16.129699584087327,10.531196919733919 -Llama3RotaryEmbedding,1.6660966986441785,4.597028827697571,2.9309321290533923 -LogitsProcessor,3.851259651293784,1.2945466160636456,-2.556713035230138 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",36.71509967192245,33.61447432909596,-3.10062534282649 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.863807294141058,9.318807340064149,-0.5449999540769088 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,3.166628119696359,3.3362354235618112,0.16960730386545242 -RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,,0.05345419748630216, -"RowParallelLinear(weight=bfloat16[4096, 14336])",25.70113205093823,20.57625382758369,-5.124878223354539 -"RowParallelLinear(weight=bfloat16[4096, 4096])",7.7274924679659325,7.381475875001642,-0.34601659296429066 -Sampler,1.3505386801824866,1.213384558039527,-0.13715412214295952 -SiluAndMul,4.202015117001586,2.4459984784793796,-1.756016638522206 -others,0.050253827955756426,0.03864094283900527,-0.011612885116751154 -vocab_embed_ops,0.1071737559047699,, diff --git a/profiling_bs1_pl256/prefill_comparison.pkl b/profiling_bs1_pl256/prefill_comparison.pkl deleted file mode 100644 index 55083015d74d3d02f62a91321eb2890919cfd27e..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl256/prefill_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f2216588373691186e8349a851050b2e05a3b5836258c4979258e026a51f5639 -size 1526 diff --git a/profiling_bs1_pl512/H100decode.pkl b/profiling_bs1_pl512/H100decode.pkl deleted file mode 100644 index d67d153aa4023c02be7bce99f4cc532f0fdb8b9f..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/H100decode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:66e6aaa974ddb2f28c76a38486992ce8a29bb8e6d5e84117e018027bc7ba731f -size 1133 diff --git a/profiling_bs1_pl512/H100decode_steps.png b/profiling_bs1_pl512/H100decode_steps.png deleted file mode 100644 index 44834a4b2424825cfcff5a628b621a0bb8390904..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/H100decode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:3e6da29952814600e7d68df0dd43169ae1471576c832d160cdbf6f6a57cb8dd4 -size 71877 diff --git a/profiling_bs1_pl512/H100prefill.pkl b/profiling_bs1_pl512/H100prefill.pkl deleted file mode 100644 index f87b90a3971bd8e34f64073ab4c5d4c2ba3668d3..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/H100prefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:628e59353e6bfc05eb3813fbe8f47c4cccc31bded22dac4bc82c9a1d35233a47 -size 1149 diff --git a/profiling_bs1_pl512/H100prefill.png b/profiling_bs1_pl512/H100prefill.png deleted file mode 100644 index a737fb69b8a96a1bebca4abb6c186d9ff9553406..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/H100prefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:d3d7ca084f3d67a4e4117137e80480df9d08dab54e721fee6141c6cd05040e81 -size 65519 diff --git a/profiling_bs1_pl512/MI300xdecode.pkl b/profiling_bs1_pl512/MI300xdecode.pkl deleted file mode 100644 index 1f05f77d625a3ea4c69fa1c8cba87167fa78db46..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/MI300xdecode.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:dcb0caa2b6400abd11b3b68edc44d3ffbd868cb6bd8d766b2d09c517718c7559 -size 1133 diff --git a/profiling_bs1_pl512/MI300xdecode_steps.png b/profiling_bs1_pl512/MI300xdecode_steps.png deleted file mode 100644 index be36c762c184c88d12f80e1276f6572240d2153d..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/MI300xdecode_steps.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:390b8a54cc135feb3932495d67fd161e9ae24939206c7b9de22e05fd20fc9d96 -size 71745 diff --git a/profiling_bs1_pl512/MI300xprefill.pkl b/profiling_bs1_pl512/MI300xprefill.pkl deleted file mode 100644 index ca5d14f960263503609cfe635ed70bd107b515b3..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/MI300xprefill.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:a1a7fbc3540a1fbec544d375af152647826e32224e9b6e3e8985f487b16682f3 -size 1149 diff --git a/profiling_bs1_pl512/MI300xprefill.png b/profiling_bs1_pl512/MI300xprefill.png deleted file mode 100644 index 03db8f911dba258256cb4b2fdbd23635d417079e..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/MI300xprefill.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:e004a2eacc2484f6f61981c2c457b60b61774f79c89f9031b80af605faa281fe -size 65561 diff --git a/profiling_bs1_pl512/decode_comparison.csv b/profiling_bs1_pl512/decode_comparison.csv deleted file mode 100644 index 2fe5191c5025e5ab69709dfb859da65b3e04a669..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/decode_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,decode_1_H100,decode_1_MI300x,abs_diff -Attention,8.041442577810386,7.51965041513902,-0.5217921626713657 -Llama3RotaryEmbedding,1.7114829769898674,2.1461022733057886,0.4346192963159212 -LogitsProcessor,5.075120442966566,4.613895956236813,-0.461224486729753 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",37.49091482307884,35.72644823677663,-1.7644665863022055 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",9.759952642396774,9.718551506926916,-0.04140113546985802 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,3.1546948322856263,5.2684228601039464,2.11372802781832 -RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,0.048152350765331656,0.0777636673800397,0.029611316614708047 -"RowParallelLinear(weight=bfloat16[4096, 14336])",20.333820262292992,19.236247071178862,-1.0975731911141295 -"RowParallelLinear(weight=bfloat16[4096, 4096])",8.567875165030642,7.745698637010221,-0.822176528020421 -Sampler,1.768999908107316,3.2032332890777298,1.4342333809704138 -SiluAndMul,4.006886253657995,4.679518344615333,0.6726320909573378 -others,0.04065776461769357,0.06446774224869432,0.023809977631000755 diff --git a/profiling_bs1_pl512/decode_comparison.pkl b/profiling_bs1_pl512/decode_comparison.pkl deleted file mode 100644 index a07bc5df947ad65069d223ab8af03413a79bb3d9..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/decode_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:93d3e103fed3ff35648a2fb1b90c7c5bb891be03f51ffbcdf5cca3c32a73d8e2 -size 1520 diff --git a/profiling_bs1_pl512/prefill_comparison.csv b/profiling_bs1_pl512/prefill_comparison.csv deleted file mode 100644 index cfb84bc97e9d6da1ceaeac5457e0578c519dc19f..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/prefill_comparison.csv +++ /dev/null @@ -1,13 +0,0 @@ -name,prefill_H100,prefill_MI300x,abs_diff -Attention,4.6992201900505775,17.07305885222289,12.37383866217231 -Llama3RotaryEmbedding,1.654183418481141,4.420474259917995,2.766290841436854 -LogitsProcessor,2.6814815672764203,0.8820516596336058,-1.7994299076428146 -"MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",41.146126621848175,33.892629321650936,-7.253497300197239 -"QKVParallelLinear(weight=bfloat16[6144, 4096])",10.96928445819218,9.449344935966167,-1.5199395222260126 -RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,3.354452265386841,3.4253512312074617,0.0708989658206205 -"RowParallelLinear(weight=bfloat16[4096, 14336])",20.778851360320235,20.81044052883708,0.031589168516845234 -"RowParallelLinear(weight=bfloat16[4096, 4096])",8.061780077104702,5.851459916155487,-2.210320160949215 -Sampler,0.8995345280289777,0.8986769869355334,-0.0008575410934442695 -SiluAndMul,5.5758567828256185,3.0850363771070417,-2.4908204057185768 -others,0.05437592877606764,0.05176087286649762,-0.0026150559095700146 -vocab_embed_ops,0.12485280170906228,0.1597150574992982,0.03486225579023593 diff --git a/profiling_bs1_pl512/prefill_comparison.pkl b/profiling_bs1_pl512/prefill_comparison.pkl deleted file mode 100644 index ee4b1bbfc60daa7d6c4f615f76822555c9531594..0000000000000000000000000000000000000000 --- a/profiling_bs1_pl512/prefill_comparison.pkl +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:8c179554ee4421afe4aedcf6795810b3dd1e9385532ab75ae8fb67afdf8e1286 -size 1483 diff --git a/vocab_embed_ops_prefill_abs_diff.png b/vocab_embed_ops_prefill_abs_diff.png deleted file mode 100644 index 91497d12b6a797717bb8cbc96a6f334d392e64dc..0000000000000000000000000000000000000000 --- a/vocab_embed_ops_prefill_abs_diff.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:c4c594ce4dee15c85cf6e43376555475a2699688b84587448f1d606a88bf73bd -size 40950 diff --git a/vocab_embed_ops_prefill_prefill_H100.png b/vocab_embed_ops_prefill_prefill_H100.png deleted file mode 100644 index a288835e53a5f99f4ff27009d35adeb0db508af0..0000000000000000000000000000000000000000 --- a/vocab_embed_ops_prefill_prefill_H100.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:6aa44ab4a8b85f5f37f392bc1edc79249db356724e69f6c14124f719195d721a -size 38894 diff --git a/vocab_embed_ops_prefill_prefill_MI300x.png b/vocab_embed_ops_prefill_prefill_MI300x.png deleted file mode 100644 index 6f6347cbfc9458df8e004033d67fa1a8dbfb430b..0000000000000000000000000000000000000000 --- a/vocab_embed_ops_prefill_prefill_MI300x.png +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f1fa766560af6edd864e283d4e283c0b07539d04fa86ac0735e6e527bb503d9c -size 37439