sskj-h3/sglang-profile/results/h3-sdpa-kernel-matrix-20260825-run1/jobs/gpu1-F3-cudnn-model_strided.trace.json

585 lines
18 KiB
JSON

{
"schemaVersion": 1,
"deviceProperties": [
{
"id": 0, "name": "NVIDIA RTX 6000D", "totalGlobalMem": 89171361792,
"computeMajor": 12, "computeMinor": 0,
"maxThreadsPerBlock": 1024, "maxThreadsPerMultiprocessor": 1536,
"regsPerBlock": 65536, "warpSize": 32,
"sharedMemPerBlock": 49152, "numSms": 156
, "regsPerMultiprocessor": 65536, "sharedMemPerBlockOptin": 101376, "sharedMemPerMultiprocessor": 102400
}
],
"trace_id": "3A82191C94E6456892B317B9A4FE7C39",
"record_shapes": 1,
"cuda_driver_version": 13000,
"cuda_runtime_version": 13000,
"cupti_version": 130001,
"displayTimeUnit": "ms",
"baseTimeNanoseconds": 1782967788000000000,
"traceEvents": [
{
"ph": "X", "cat": "cpu_op", "name": "aten::scaled_dot_product_attention", "pid": 2815986, "tid": 2815986,
"ts": 4656932412357.139, "dur": 3881.981,
"args": {
"External id": 1,"Sequence number": 0, "Fwd thread id": 0, "Record function id": 0, "Concrete Inputs": ["", "", "", "", "0.", "False", "", "False"], "Input type": ["c10::BFloat16", "c10::BFloat16", "c10::BFloat16", "", "Scalar", "Scalar", "", "Scalar"], "Input Strides": [[449197056, 128, 10752, 1], [449197056, 128, 10752, 1], [449197056, 128, 10752, 1], [], [], [], [], []], "Input Dims": [[1, 28, 41778, 128], [1, 28, 41778, 128], [1, 28, 41778, 128], [], [], [], [], []], "Ev Idx": 0
}
},
{
"ph": "X", "cat": "cpu_op", "name": "aten::_scaled_dot_product_cudnn_attention", "pid": 2815986, "tid": 2815986,
"ts": 4656932412474.871, "dur": 3756.884,
"args": {
"External id": 2,"Sequence number": 0, "Fwd thread id": 0, "Record function id": 0, "Concrete Inputs": ["", "", "", "", "False", "0.", "False", "False", ""], "Input type": ["c10::BFloat16", "c10::BFloat16", "c10::BFloat16", "", "Scalar", "Scalar", "Scalar", "Scalar", ""], "Input Strides": [[449197056, 128, 10752, 1], [449197056, 128, 10752, 1], [449197056, 128, 10752, 1], [], [], [], [], [], []], "Input Dims": [[1, 28, 41778, 128], [1, 28, 41778, 128], [1, 28, 41778, 128], [], [], [], [], [], []], "Ev Idx": 1
}
},
{
"ph": "X", "cat": "cpu_op", "name": "aten::_cudnn_attention_forward", "pid": 2815986, "tid": 2815986,
"ts": 4656932412513.794, "dur": 3713.081,
"args": {
"External id": 3,"Record function id": 0, "Concrete Inputs": ["", "", "", "", "", "", "41778", "41778", "False", "0.", "False", "False", ""], "Input type": ["c10::BFloat16", "c10::BFloat16", "c10::BFloat16", "", "", "", "Scalar", "Scalar", "Scalar", "Scalar", "Scalar", "Scalar", ""], "Input Strides": [[449197056, 128, 10752, 1], [449197056, 128, 10752, 1], [449197056, 128, 10752, 1], [], [], [], [], [], [], [], [], [], []], "Input Dims": [[1, 28, 41778, 128], [1, 28, 41778, 128], [1, 28, 41778, 128], [], [], [], [], [], [], [], [], [], []], "Ev Idx": 2
}
},
{
"ph": "X", "cat": "cpu_op", "name": "aten::empty", "pid": 2815986, "tid": 2815986,
"ts": 4656932412537.318, "dur": 3154.100,
"args": {
"External id": 4,"Record function id": 0, "Concrete Inputs": ["[]", "4", "", "", "", ""], "Input type": ["ScalarList", "Scalar", "", "", "", ""], "Input Strides": [[], [], [], [], [], []], "Input Dims": [[], [], [], [], [], []], "Ev Idx": 3
}
},
{
"ph": "X", "cat": "cpu_op", "name": "aten::empty", "pid": 2815986, "tid": 2815986,
"ts": 4656932415706.135, "dur": 13.724,
"args": {
"External id": 5,"Record function id": 0, "Concrete Inputs": ["[]", "4", "", "", "", ""], "Input type": ["ScalarList", "Scalar", "", "", "", ""], "Input Strides": [[], [], [], [], [], []], "Input Dims": [[], [], [], [], [], []], "Ev Idx": 4
}
},
{
"ph": "X", "cat": "cpu_op", "name": "aten::empty_like", "pid": 2815986, "tid": 2815986,
"ts": 4656932415750.463, "dur": 275.371,
"args": {
"External id": 6,"Record function id": 0, "Concrete Inputs": ["", "", "", "", "", ""], "Input type": ["c10::BFloat16", "", "", "", "", ""], "Input Strides": [[449197056, 128, 10752, 1], [], [], [], [], []], "Input Dims": [[1, 28, 41778, 128], [], [], [], [], []], "Ev Idx": 5
}
},
{
"ph": "X", "cat": "cpu_op", "name": "aten::empty_strided", "pid": 2815986, "tid": 2815986,
"ts": 4656932415763.060, "dur": 260.909,
"args": {
"External id": 7,"Record function id": 0, "Concrete Inputs": ["[1, 28, 41778, 128]", "[149732352, 128, 3584, 1]", "15", "0", "", ""], "Input type": ["ScalarList", "ScalarList", "Scalar", "Scalar", "", ""], "Input Strides": [[], [], [], [], [], []], "Input Dims": [[], [], [], [], [], []], "Ev Idx": 6
}
},
{
"ph": "X", "cat": "overhead", "name": "Activity Buffer Request", "pid": -1, "tid": 0,
"ts": 4656932412570.033, "dur": 2712.932
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaStreamIsCapturing", "pid": 2815986, "tid": 2815986,
"ts": 4656932415300.200, "dur": 20.498,
"args": {
"External id": 4, "cbid": 317, "correlation": 13
}
},
{
"ph": "f", "id": 13, "pid": 2815986, "tid": 2815986, "ts": 4656932415300.200,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaMalloc", "pid": 2815986, "tid": 2815986,
"ts": 4656932415322.211, "dur": 336.776,
"args": {
"External id": 4, "cbid": 20, "correlation": 14
}
},
{
"ph": "f", "id": 14, "pid": 2815986, "tid": 2815986, "ts": 4656932415322.211,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaStreamIsCapturing", "pid": 2815986, "tid": 2815986,
"ts": 4656932415727.396, "dur": 1.400,
"args": {
"External id": 3, "cbid": 317, "correlation": 29
}
},
{
"ph": "f", "id": 29, "pid": 2815986, "tid": 2815986, "ts": 4656932415727.396,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaStreamIsCapturing", "pid": 2815986, "tid": 2815986,
"ts": 4656932415774.786, "dur": 0.414,
"args": {
"External id": 7, "cbid": 317, "correlation": 35
}
},
{
"ph": "f", "id": 35, "pid": 2815986, "tid": 2815986, "ts": 4656932415774.786,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaMalloc", "pid": 2815986, "tid": 2815986,
"ts": 4656932415775.690, "dur": 238.796,
"args": {
"External id": 7, "cbid": 20, "correlation": 36
}
},
{
"ph": "f", "id": 36, "pid": 2815986, "tid": 2815986, "ts": 4656932415775.690,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaEventRecord", "pid": 2815986, "tid": 2815986,
"ts": 4656932416045.919, "dur": 10.532,
"args": {
"External id": 3, "cbid": 135, "correlation": 44
}
},
{
"ph": "f", "id": 44, "pid": 2815986, "tid": 2815986, "ts": 4656932416045.919,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaStreamGetPriority", "pid": 2815986, "tid": 2815986,
"ts": 4656932416057.952, "dur": 3.434,
"args": {
"External id": 3, "cbid": 203, "correlation": 45
}
},
{
"ph": "f", "id": 45, "pid": 2815986, "tid": 2815986, "ts": 4656932416057.952,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaDeviceGetStreamPriorityRange", "pid": 2815986, "tid": 2815986,
"ts": 4656932416062.388, "dur": 1.140,
"args": {
"External id": 3, "cbid": 205, "correlation": 46
}
},
{
"ph": "f", "id": 46, "pid": 2815986, "tid": 2815986, "ts": 4656932416062.388,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "kernel", "name": "cudnn_generated_fort_native_sdpa_sm120_flash_fprop_f16_knob_3_64x64x128_4x1x1_cga1x1x1_kernel0_0", "pid": 0, "tid": 7,
"ts": 4656932416206.287, "dur": 175094.204,
"args": {
"External id": 3, "queued": 0, "device": 0, "context": 1, "stream": 7, "correlation": 67, "registers per thread": 168, "shared memory": 101376, "blocks per SM": 58.692307, "warps per SM": 528.230774, "grid": [327, 28, 1], "block": [288, 1, 1], "est. achieved occupancy %": 0
}
},
{
"ph": "f", "id": 67, "pid": 0, "tid": 7, "ts": 4656932416206.287,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_driver", "name": "cuLaunchKernelEx", "pid": 2815986, "tid": 2815986,
"ts": 4656932416149.847, "dur": 56.386,
"args": {
"External id": 3, "cbid": 652, "correlation": 67
}
},
{
"ph": "s", "id": 67, "pid": 2815986, "tid": 2815986, "ts": 4656932416149.847,
"cat": "ac2g", "name": "ac2g"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaGetErrorString", "pid": 2815986, "tid": 2815986,
"ts": 4656932416214.438, "dur": 2.496,
"args": {
"External id": 3, "cbid": 12, "correlation": 69
}
},
{
"ph": "f", "id": 69, "pid": 2815986, "tid": 2815986, "ts": 4656932416214.438,
"cat": "ac2g", "name": "ac2g", "bp": "e"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaDeviceSynchronize", "pid": 2815986, "tid": 2815986,
"ts": 4656932416363.511, "dur": 174951.538,
"args": {
"cbid": 165, "correlation": 76
}
},
{
"ph": "s", "id": 76, "pid": 2815986, "tid": 2815986, "ts": 4656932416363.511,
"cat": "ac2g", "name": "ac2g"
},
{
"ph": "X", "cat": "cuda_runtime", "name": "cudaDeviceSynchronize", "pid": 2815986, "tid": 2815986,
"ts": 4656932591417.152, "dur": 7.214,
"args": {
"cbid": 165, "correlation": 80
}
},
{
"ph": "s", "id": 80, "pid": 2815986, "tid": 2815986, "ts": 4656932591417.152,
"cat": "ac2g", "name": "ac2g"
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 2815986, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 2815986, "tid": 0,
"args": {
"labels": "CPU"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 2815986, "tid": 0,
"args": {
"sort_index": 2815986
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 0, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 0, "tid": 0,
"args": {
"labels": "GPU 0"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 0, "tid": 0,
"args": {
"sort_index": 5000000
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 1, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 1, "tid": 0,
"args": {
"labels": "GPU 1"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 1, "tid": 0,
"args": {
"sort_index": 5000001
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 2, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 2, "tid": 0,
"args": {
"labels": "GPU 2"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 2, "tid": 0,
"args": {
"sort_index": 5000002
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 3, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 3, "tid": 0,
"args": {
"labels": "GPU 3"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 3, "tid": 0,
"args": {
"sort_index": 5000003
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 4, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 4, "tid": 0,
"args": {
"labels": "GPU 4"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 4, "tid": 0,
"args": {
"sort_index": 5000004
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 5, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 5, "tid": 0,
"args": {
"labels": "GPU 5"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 5, "tid": 0,
"args": {
"sort_index": 5000005
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 6, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 6, "tid": 0,
"args": {
"labels": "GPU 6"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 6, "tid": 0,
"args": {
"sort_index": 5000006
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 7, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 7, "tid": 0,
"args": {
"labels": "GPU 7"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 7, "tid": 0,
"args": {
"sort_index": 5000007
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 8, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 8, "tid": 0,
"args": {
"labels": "GPU 8"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 8, "tid": 0,
"args": {
"sort_index": 5000008
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 9, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 9, "tid": 0,
"args": {
"labels": "GPU 9"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 9, "tid": 0,
"args": {
"sort_index": 5000009
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 10, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 10, "tid": 0,
"args": {
"labels": "GPU 10"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 10, "tid": 0,
"args": {
"sort_index": 5000010
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 11, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 11, "tid": 0,
"args": {
"labels": "GPU 11"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 11, "tid": 0,
"args": {
"sort_index": 5000011
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 12, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 12, "tid": 0,
"args": {
"labels": "GPU 12"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 12, "tid": 0,
"args": {
"sort_index": 5000012
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 13, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 13, "tid": 0,
"args": {
"labels": "GPU 13"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 13, "tid": 0,
"args": {
"sort_index": 5000013
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 14, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 14, "tid": 0,
"args": {
"labels": "GPU 14"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 14, "tid": 0,
"args": {
"sort_index": 5000014
}
},
{
"name": "process_name", "ph": "M", "ts": 4656932412164.433, "pid": 15, "tid": 0,
"args": {
"name": "python"
}
},
{
"name": "process_labels", "ph": "M", "ts": 4656932412164.433, "pid": 15, "tid": 0,
"args": {
"labels": "GPU 15"
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 15, "tid": 0,
"args": {
"sort_index": 5000015
}
},
{
"name": "thread_name", "ph": "M", "ts": 4656932412164.433, "pid": 0, "tid": 7,
"args": {
"name": "stream 7 "
}
},
{
"name": "thread_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 0, "tid": 7,
"args": {
"sort_index": 7
}
},
{
"name": "thread_name", "ph": "M", "ts": 4656932412164.433, "pid": 2815986, "tid": 2815986,
"args": {
"name": "thread 2815986 (python)"
}
},
{
"name": "thread_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 2815986, "tid": 2815986,
"args": {
"sort_index": 2815986
}
},
{
"name": "thread_name", "ph": "M", "ts": 4656932412164.433, "pid": 2815986, "tid": 2815986,
"args": {
"name": "thread 2815986 (python)"
}
},
{
"name": "thread_sort_index", "ph": "M", "ts": 4656932412164.433, "pid": 2815986, "tid": 2815986,
"args": {
"sort_index": 2815986
}
},
{
"ph": "X", "cat": "Trace", "ts": 4656932412098.210, "dur": 179344.981,
"pid": "Spans", "tid": "PyTorch Profiler",
"name": "PyTorch Profiler (0)",
"args": {
"Op count": 0
}
},
{
"name": "process_sort_index", "ph": "M", "ts": 4656932412098.210,
"pid": "Spans", "tid": 0,
"args": {
"sort_index": 536870912
}
},
{
"name": "Iteration Start: PyTorch Profiler", "ph": "i", "s": "g",
"pid": "Traces", "tid": "Trace PyTorch Profiler", "ts": 4656932412098.210
},
{
"name": "Record Window End", "ph": "i", "s": "g",
"pid": "", "tid": "", "ts": 4656932591714.706
}
],
"traceName": "/data/wxy/profile_results/h3-sdpa-kernel-matrix-20260825-run1/jobs/gpu1-F3-cudnn-model_strided.trace.json"
}