{ "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": "DEBEC15BFCAD408F87AD9537A89A82E8", "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": 2815987, "tid": 2815987, "ts": 4656950376677.283, "dur": 2262.700, "args": { "External id": 513,"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": [[149732352, 5347584, 128, 1], [149732352, 5347584, 128, 1], [149732352, 5347584, 128, 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_efficient_attention", "pid": 2815987, "tid": 2815987, "ts": 4656950376715.840, "dur": 2221.283, "args": { "External id": 514,"Sequence number": 0, "Fwd thread id": 0, "Record function id": 0, "Concrete Inputs": ["", "", "", "", "False", "0.", "False", ""], "Input type": ["c10::BFloat16", "c10::BFloat16", "c10::BFloat16", "", "Scalar", "Scalar", "Scalar", ""], "Input Strides": [[149732352, 5347584, 128, 1], [149732352, 5347584, 128, 1], [149732352, 5347584, 128, 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::transpose", "pid": 2815987, "tid": 2815987, "ts": 4656950376728.681, "dur": 19.108, "args": { "External id": 515,"Record function id": 0, "Concrete Inputs": ["", "1", "2"], "Input type": ["c10::BFloat16", "Scalar", "Scalar"], "Input Strides": [[149732352, 5347584, 128, 1], [], []], "Input Dims": [[1, 28, 41778, 128], [], []], "Ev Idx": 2 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::as_strided", "pid": 2815987, "tid": 2815987, "ts": 4656950376741.969, "dur": 3.927, "args": { "External id": 516,"Record function id": 0, "Concrete Inputs": ["", "[1, 41778, 28, 128]", "[149732352, 128, 5347584, 1]", ""], "Input type": ["c10::BFloat16", "ScalarList", "ScalarList", ""], "Input Strides": [[149732352, 5347584, 128, 1], [], [], []], "Input Dims": [[1, 28, 41778, 128], [], [], []], "Ev Idx": 3 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::transpose", "pid": 2815987, "tid": 2815987, "ts": 4656950376748.669, "dur": 2.029, "args": { "External id": 517,"Record function id": 0, "Concrete Inputs": ["", "1", "2"], "Input type": ["c10::BFloat16", "Scalar", "Scalar"], "Input Strides": [[149732352, 5347584, 128, 1], [], []], "Input Dims": [[1, 28, 41778, 128], [], []], "Ev Idx": 4 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::as_strided", "pid": 2815987, "tid": 2815987, "ts": 4656950376749.950, "dur": 0.449, "args": { "External id": 518,"Record function id": 0, "Concrete Inputs": ["", "[1, 41778, 28, 128]", "[149732352, 128, 5347584, 1]", ""], "Input type": ["c10::BFloat16", "ScalarList", "ScalarList", ""], "Input Strides": [[149732352, 5347584, 128, 1], [], [], []], "Input Dims": [[1, 28, 41778, 128], [], [], []], "Ev Idx": 5 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::transpose", "pid": 2815987, "tid": 2815987, "ts": 4656950376751.474, "dur": 1.500, "args": { "External id": 519,"Record function id": 0, "Concrete Inputs": ["", "1", "2"], "Input type": ["c10::BFloat16", "Scalar", "Scalar"], "Input Strides": [[149732352, 5347584, 128, 1], [], []], "Input Dims": [[1, 28, 41778, 128], [], []], "Ev Idx": 6 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::as_strided", "pid": 2815987, "tid": 2815987, "ts": 4656950376752.374, "dur": 0.336, "args": { "External id": 520,"Record function id": 0, "Concrete Inputs": ["", "[1, 41778, 28, 128]", "[149732352, 128, 5347584, 1]", ""], "Input type": ["c10::BFloat16", "ScalarList", "ScalarList", ""], "Input Strides": [[149732352, 5347584, 128, 1], [], [], []], "Input Dims": [[1, 28, 41778, 128], [], [], []], "Ev Idx": 7 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::_efficient_attention_forward", "pid": 2815987, "tid": 2815987, "ts": 4656950376758.296, "dur": 2165.548, "args": { "External id": 521,"Record function id": 0, "Concrete Inputs": ["", "", "", "", "", "", "", "", "0.", "0", "False", "", "", ""], "Input type": ["c10::BFloat16", "c10::BFloat16", "c10::BFloat16", "", "", "", "", "", "Scalar", "Scalar", "Scalar", "", "", ""], "Input Strides": [[149732352, 128, 5347584, 1], [149732352, 128, 5347584, 1], [149732352, 128, 5347584, 1], [], [], [], [], [], [], [], [], [], [], []], "Input Dims": [[1, 41778, 28, 128], [1, 41778, 28, 128], [1, 41778, 28, 128], [], [], [], [], [], [], [], [], [], [], []], "Ev Idx": 8 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::empty", "pid": 2815987, "tid": 2815987, "ts": 4656950378522.122, "dur": 10.637, "args": { "External id": 522,"Record function id": 0, "Concrete Inputs": ["[]", "4", "", "", "", ""], "Input type": ["ScalarList", "Scalar", "", "", "", ""], "Input Strides": [[], [], [], [], [], []], "Input Dims": [[], [], [], [], [], []], "Ev Idx": 9 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::empty", "pid": 2815987, "tid": 2815987, "ts": 4656950378534.826, "dur": 0.846, "args": { "External id": 523,"Record function id": 0, "Concrete Inputs": ["[]", "4", "", "", "", ""], "Input type": ["ScalarList", "Scalar", "", "", "", ""], "Input Strides": [[], [], [], [], [], []], "Input Dims": [[], [], [], [], [], []], "Ev Idx": 10 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::empty", "pid": 2815987, "tid": 2815987, "ts": 4656950378538.945, "dur": 310.038, "args": { "External id": 524,"Record function id": 0, "Concrete Inputs": ["[1, 41778, 28, 128]", "15", "0", "", "", ""], "Input type": ["ScalarList", "Scalar", "Scalar", "", "", ""], "Input Strides": [[], [], [], [], [], []], "Input Dims": [[], [], [], [], [], []], "Ev Idx": 11 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::empty", "pid": 2815987, "tid": 2815987, "ts": 4656950378852.253, "dur": 4.103, "args": { "External id": 525,"Record function id": 0, "Concrete Inputs": ["[1, 28, 0]", "6", "0", "", "", ""], "Input type": ["ScalarList", "Scalar", "Scalar", "", "", ""], "Input Strides": [[], [], [], [], [], []], "Input Dims": [[], [], [], [], [], []], "Ev Idx": 12 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::transpose", "pid": 2815987, "tid": 2815987, "ts": 4656950378926.585, "dur": 7.109, "args": { "External id": 526,"Record function id": 0, "Concrete Inputs": ["", "1", "2"], "Input type": ["c10::BFloat16", "Scalar", "Scalar"], "Input Strides": [[149732352, 3584, 128, 1], [], []], "Input Dims": [[1, 41778, 28, 128], [], []], "Ev Idx": 13 } }, { "ph": "X", "cat": "cpu_op", "name": "aten::as_strided", "pid": 2815987, "tid": 2815987, "ts": 4656950378929.901, "dur": 3.334, "args": { "External id": 527,"Record function id": 0, "Concrete Inputs": ["", "[1, 28, 41778, 128]", "[149732352, 128, 3584, 1]", ""], "Input type": ["c10::BFloat16", "ScalarList", "ScalarList", ""], "Input Strides": [[149732352, 3584, 128, 1], [], [], []], "Input Dims": [[1, 41778, 28, 128], [], [], []], "Ev Idx": 14 } }, { "ph": "X", "cat": "overhead", "name": "Activity Buffer Request", "pid": -1, "tid": 0, "ts": 4656950376768.013, "dur": 1737.388 }, { "ph": "X", "cat": "cuda_runtime", "name": "cudaStreamIsCapturing", "pid": 2815987, "tid": 2815987, "ts": 4656950378507.564, "dur": 4.866, "args": { "External id": 521, "cbid": 317, "correlation": 1940 } }, { "ph": "f", "id": 1940, "pid": 2815987, "tid": 2815987, "ts": 4656950378507.564, "cat": "ac2g", "name": "ac2g", "bp": "e" }, { "ph": "X", "cat": "cuda_runtime", "name": "cudaStreamIsCapturing", "pid": 2815987, "tid": 2815987, "ts": 4656950378556.624, "dur": 0.332, "args": { "External id": 524, "cbid": 317, "correlation": 1946 } }, { "ph": "f", "id": 1946, "pid": 2815987, "tid": 2815987, "ts": 4656950378556.624, "cat": "ac2g", "name": "ac2g", "bp": "e" }, { "ph": "X", "cat": "cuda_runtime", "name": "cudaMalloc", "pid": 2815987, "tid": 2815987, "ts": 4656950378557.680, "dur": 283.837, "args": { "External id": 524, "cbid": 20, "correlation": 1947 } }, { "ph": "f", "id": 1947, "pid": 2815987, "tid": 2815987, "ts": 4656950378557.680, "cat": "ac2g", "name": "ac2g", "bp": "e" }, { "ph": "X", "cat": "cuda_runtime", "name": "cudaFuncSetAttribute", "pid": 2815987, "tid": 2815987, "ts": 4656950378860.949, "dur": 12.729, "args": { "External id": 521, "cbid": 273, "correlation": 1959 } }, { "ph": "f", "id": 1959, "pid": 2815987, "tid": 2815987, "ts": 4656950378860.949, "cat": "ac2g", "name": "ac2g", "bp": "e" }, { "ph": "X", "cat": "kernel", "name": "fmha_cutlassF_bf16_aligned_64x128_rf_sm80(PyTorchMemEffAttention::AttentionKernel::Params)", "pid": 0, "tid": 7, "ts": 4656950378919.588, "dur": 353623.030, "args": { "External id": 521, "queued": 0, "device": 0, "context": 1, "stream": 7, "correlation": 1961, "registers per thread": 126, "shared memory": 52224, "blocks per SM": 117.205132, "warps per SM": 937.641052, "grid": [653, 28, 1], "block": [32, 8, 1], "est. achieved occupancy %": 0 } }, { "ph": "f", "id": 1961, "pid": 0, "tid": 7, "ts": 4656950378919.588, "cat": "ac2g", "name": "ac2g", "bp": "e" }, { "ph": "X", "cat": "cuda_runtime", "name": "cudaLaunchKernel", "pid": 2815987, "tid": 2815987, "ts": 4656950378878.117, "dur": 41.886, "args": { "External id": 521, "cbid": 211, "correlation": 1961 } }, { "ph": "s", "id": 1961, "pid": 2815987, "tid": 2815987, "ts": 4656950378878.117, "cat": "ac2g", "name": "ac2g" }, { "ph": "X", "cat": "cuda_runtime", "name": "cudaDeviceSynchronize", "pid": 2815987, "tid": 2815987, "ts": 4656950379015.485, "dur": 353545.114, "args": { "cbid": 165, "correlation": 1971 } }, { "ph": "s", "id": 1971, "pid": 2815987, "tid": 2815987, "ts": 4656950379015.485, "cat": "ac2g", "name": "ac2g" }, { "ph": "X", "cat": "cuda_runtime", "name": "cudaDeviceSynchronize", "pid": 2815987, "tid": 2815987, "ts": 4656950732732.150, "dur": 12.304, "args": { "cbid": 165, "correlation": 1975 } }, { "ph": "s", "id": 1975, "pid": 2815987, "tid": 2815987, "ts": 4656950732732.150, "cat": "ac2g", "name": "ac2g" }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 2815987, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 2815987, "tid": 0, "args": { "labels": "CPU" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 2815987, "tid": 0, "args": { "sort_index": 2815987 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 0, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 0, "tid": 0, "args": { "labels": "GPU 0" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 0, "tid": 0, "args": { "sort_index": 5000000 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 1, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 1, "tid": 0, "args": { "labels": "GPU 1" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 1, "tid": 0, "args": { "sort_index": 5000001 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 2, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 2, "tid": 0, "args": { "labels": "GPU 2" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 2, "tid": 0, "args": { "sort_index": 5000002 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 3, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 3, "tid": 0, "args": { "labels": "GPU 3" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 3, "tid": 0, "args": { "sort_index": 5000003 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 4, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 4, "tid": 0, "args": { "labels": "GPU 4" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 4, "tid": 0, "args": { "sort_index": 5000004 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 5, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 5, "tid": 0, "args": { "labels": "GPU 5" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 5, "tid": 0, "args": { "sort_index": 5000005 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 6, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 6, "tid": 0, "args": { "labels": "GPU 6" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 6, "tid": 0, "args": { "sort_index": 5000006 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 7, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 7, "tid": 0, "args": { "labels": "GPU 7" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 7, "tid": 0, "args": { "sort_index": 5000007 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 8, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 8, "tid": 0, "args": { "labels": "GPU 8" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 8, "tid": 0, "args": { "sort_index": 5000008 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 9, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 9, "tid": 0, "args": { "labels": "GPU 9" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 9, "tid": 0, "args": { "sort_index": 5000009 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 10, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 10, "tid": 0, "args": { "labels": "GPU 10" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 10, "tid": 0, "args": { "sort_index": 5000010 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 11, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 11, "tid": 0, "args": { "labels": "GPU 11" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 11, "tid": 0, "args": { "sort_index": 5000011 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 12, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 12, "tid": 0, "args": { "labels": "GPU 12" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 12, "tid": 0, "args": { "sort_index": 5000012 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 13, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 13, "tid": 0, "args": { "labels": "GPU 13" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 13, "tid": 0, "args": { "sort_index": 5000013 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 14, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 14, "tid": 0, "args": { "labels": "GPU 14" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 14, "tid": 0, "args": { "sort_index": 5000014 } }, { "name": "process_name", "ph": "M", "ts": 4656950376420.695, "pid": 15, "tid": 0, "args": { "name": "python" } }, { "name": "process_labels", "ph": "M", "ts": 4656950376420.695, "pid": 15, "tid": 0, "args": { "labels": "GPU 15" } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 15, "tid": 0, "args": { "sort_index": 5000015 } }, { "name": "thread_name", "ph": "M", "ts": 4656950376420.695, "pid": 0, "tid": 7, "args": { "name": "stream 7 " } }, { "name": "thread_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 0, "tid": 7, "args": { "sort_index": 7 } }, { "name": "thread_name", "ph": "M", "ts": 4656950376420.695, "pid": 2815987, "tid": 2815987, "args": { "name": "thread 2815987 (python)" } }, { "name": "thread_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 2815987, "tid": 2815987, "args": { "sort_index": 2815987 } }, { "name": "thread_name", "ph": "M", "ts": 4656950376420.695, "pid": 2815987, "tid": 2815987, "args": { "name": "thread 2815987 (python)" } }, { "name": "thread_sort_index", "ph": "M", "ts": 4656950376420.695, "pid": 2815987, "tid": 2815987, "args": { "sort_index": 2815987 } }, { "ph": "X", "cat": "Trace", "ts": 4656950376371.494, "dur": 356401.274, "pid": "Spans", "tid": "PyTorch Profiler", "name": "PyTorch Profiler (0)", "args": { "Op count": 0 } }, { "name": "process_sort_index", "ph": "M", "ts": 4656950376371.494, "pid": "Spans", "tid": 0, "args": { "sort_index": 536870912 } }, { "name": "Iteration Start: PyTorch Profiler", "ph": "i", "s": "g", "pid": "Traces", "tid": "Trace PyTorch Profiler", "ts": 4656950376371.494 }, { "name": "Record Window End", "ph": "i", "s": "g", "pid": "", "tid": "", "ts": 4656950733018.972 } ], "traceName": "/data/wxy/profile_results/h3-sdpa-kernel-matrix-20260825-run1/jobs/gpu2-F3-efficient-contiguous.trace.json" }