{
  "clockHz": 1500000000,
  "tileCount": 1472,
  "sampleCount": 232260,
  "matchedSampleCount": 37376,
  "profileSpanCycles": 16746138,
  "profileSpanMs": 11.164092,
  "atOffset": null,
  "groups": [
    {
      "name": "exchange",
      "dimensions": {},
      "sampleCount": 5888,
      "tileCount": 1472,
      "phaseCount": 4,
      "phaseCycles": 94608,
      "phaseMs": 0.063072,
      "workCycles": 66939666,
      "averageActiveTiles": 707.5476281075596,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 11368.82914402174,
      "p50Cycles": 9930,
      "p95Cycles": 25608,
      "maximumCycles": 38544,
      "firstOffset": 210342,
      "lastOffset": 350916
    },
    {
      "name": "attention_softmax_d72_p768_tail_f8_sm4",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 15,
      "phaseCycles": 53034,
      "phaseMs": 0.035356,
      "workCycles": 71468532,
      "averageActiveTiles": 1347.5983708564318,
      "usefulCycles": 33499008.0,
      "estimatedWorkCycles": 71468532,
      "usefulUtilization": 0.46872388535978327,
      "workLaneOccupancy": 0.9972222222222222,
      "deviceUsefulUtilization": 0.4291111034594706,
      "meanCycles": 48551.99184782609,
      "p50Cycles": 48552,
      "p95Cycles": 48552,
      "maximumCycles": 48552,
      "firstOffset": 270666,
      "lastOffset": 323700
    },
    {
      "name": "fill_zero_u64",
      "dimensions": {},
      "sampleCount": 15296,
      "tileCount": 1472,
      "phaseCount": 45,
      "phaseCycles": 31272,
      "phaseMs": 0.020848,
      "workCycles": 5716416,
      "averageActiveTiles": 182.79662317728318,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 2911872,
      "usefulUtilization": 0.0,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": 0.0,
      "meanCycles": 373.7196652719665,
      "p50Cycles": 306,
      "p95Cycles": 726,
      "maximumCycles": 726,
      "firstOffset": 202740,
      "lastOffset": 358692
    },
    {
      "name": "gemm_f16_init_large_rows_interleaved_k80_c112_r121_r122",
      "dimensions": {},
      "sampleCount": 384,
      "tileCount": 384,
      "phaseCount": 4,
      "phaseCycles": 23310,
      "phaseMs": 0.01554,
      "workCycles": 8831232,
      "averageActiveTiles": 378.86023166023165,
      "usefulCycles": 5902848.0,
      "estimatedWorkCycles": 8831232,
      "usefulUtilization": 0.6684059483433342,
      "workLaneOccupancy": 0.9,
      "deviceUsefulUtilization": 0.17203290246768507,
      "meanCycles": 22998.0,
      "p50Cycles": 22998,
      "p95Cycles": 22998,
      "maximumCycles": 22998,
      "firstOffset": 240156,
      "lastOffset": 263466
    },
    {
      "name": "gemm_f16_init_small_rows_interleaved_k80_c112_r121_r122",
      "dimensions": {},
      "sampleCount": 384,
      "tileCount": 384,
      "phaseCount": 4,
      "phaseCycles": 23232,
      "phaseMs": 0.015488,
      "workCycles": 8815104,
      "averageActiveTiles": 379.43801652892563,
      "usefulCycles": 5854464.0,
      "estimatedWorkCycles": 8815104,
      "usefulUtilization": 0.6641400940930475,
      "workLaneOccupancy": 0.9,
      "deviceUsefulUtilization": 0.17119565217391305,
      "meanCycles": 22956.0,
      "p50Cycles": 22956,
      "p95Cycles": 22956,
      "maximumCycles": 22956,
      "firstOffset": 240174,
      "lastOffset": 263406
    },
    {
      "name": "gemm_f16_init_large_rows_interleaved_k80_c96_r121_r122",
      "dimensions": {},
      "sampleCount": 288,
      "tileCount": 288,
      "phaseCount": 3,
      "phaseCycles": 20574,
      "phaseMs": 0.013716,
      "workCycles": 5693760,
      "averageActiveTiles": 276.745406824147,
      "usefulCycles": 3702456.0,
      "estimatedWorkCycles": 5693760,
      "usefulUtilization": 0.6502655538694992,
      "workLaneOccupancy": 0.878125,
      "deviceUsefulUtilization": 0.12225407965308684,
      "meanCycles": 19770.0,
      "p50Cycles": 19770,
      "p95Cycles": 19770,
      "maximumCycles": 19770,
      "firstOffset": 239586,
      "lastOffset": 260160
    },
    {
      "name": "gemm_f16_init_small_rows_interleaved_k80_c96_r121_r122",
      "dimensions": {},
      "sampleCount": 288,
      "tileCount": 288,
      "phaseCount": 5,
      "phaseCycles": 20016,
      "phaseMs": 0.013344,
      "workCycles": 5683392,
      "averageActiveTiles": 283.9424460431655,
      "usefulCycles": 3672108.0,
      "estimatedWorkCycles": 5683392,
      "usefulUtilization": 0.6461120401337793,
      "workLaneOccupancy": 0.878125,
      "deviceUsefulUtilization": 0.12463222356897091,
      "meanCycles": 19734.0,
      "p50Cycles": 19734,
      "p95Cycles": 19734,
      "maximumCycles": 19734,
      "firstOffset": 239568,
      "lastOffset": 259584
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k128_c80_r104_r105",
      "dimensions": {},
      "sampleCount": 1152,
      "tileCount": 1152,
      "phaseCount": 14,
      "phaseCycles": 14058,
      "phaseMs": 0.009372,
      "workCycles": 13727232,
      "averageActiveTiles": 976.4711907810499,
      "usefulCycles": 8188128.0,
      "estimatedWorkCycles": 13727232,
      "usefulUtilization": 0.596487915407855,
      "workLaneOccupancy": 0.854296875,
      "deviceUsefulUtilization": 0.3956883594054445,
      "meanCycles": 11916.0,
      "p50Cycles": 11916,
      "p95Cycles": 11916,
      "maximumCycles": 11916,
      "firstOffset": 331386,
      "lastOffset": 345444
    },
    {
      "name": "gemm_f8_init_large_rows_interleaved_k128_c80_r104_r105",
      "dimensions": {},
      "sampleCount": 192,
      "tileCount": 192,
      "phaseCount": 6,
      "phaseCycles": 13578,
      "phaseMs": 0.009052,
      "workCycles": 2297088,
      "averageActiveTiles": 169.17719840919133,
      "usefulCycles": 1377810.0,
      "estimatedWorkCycles": 2297088,
      "usefulUtilization": 0.5998072342026078,
      "workLaneOccupancy": 0.854296875,
      "deviceUsefulUtilization": 0.06893594257334434,
      "meanCycles": 11964.0,
      "p50Cycles": 11964,
      "p95Cycles": 11964,
      "maximumCycles": 11964,
      "firstOffset": 332208,
      "lastOffset": 345786
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r64_p64_c32_p32_b64x16",
      "dimensions": {},
      "sampleCount": 704,
      "tileCount": 704,
      "phaseCount": 6,
      "phaseCycles": 13014,
      "phaseMs": 0.008676,
      "workCycles": 7569408,
      "averageActiveTiles": 581.6357768556938,
      "usefulCycles": 720896.0,
      "estimatedWorkCycles": 7569408,
      "usefulUtilization": 0.09523809523809523,
      "workLaneOccupancy": 1.0,
      "deviceUsefulUtilization": 0.0376317143410775,
      "meanCycles": 10752.0,
      "p50Cycles": 10752,
      "p95Cycles": 10752,
      "maximumCycles": 10752,
      "firstOffset": 210924,
      "lastOffset": 223938
    },
    {
      "name": "copy_u64",
      "dimensions": {},
      "sampleCount": 3968,
      "tileCount": 1472,
      "phaseCount": 32,
      "phaseCycles": 11892,
      "phaseMs": 0.007928,
      "workCycles": 4395264,
      "averageActiveTiles": 369.598385469223,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 1107.6774193548388,
      "p50Cycles": 1206,
      "p95Cycles": 1536,
      "maximumCycles": 1536,
      "firstOffset": 205116,
      "lastOffset": 359466
    },
    {
      "name": "cast_f16_f8",
      "dimensions": {},
      "sampleCount": 1152,
      "tileCount": 1152,
      "phaseCount": 8,
      "phaseCycles": 11880,
      "phaseMs": 0.00792,
      "workCycles": 3506994,
      "averageActiveTiles": 295.20151515151514,
      "usefulCycles": 233280.0,
      "estimatedWorkCycles": 3506994,
      "usefulUtilization": 0.06651850559196851,
      "workLaneOccupancy": 0.94921875,
      "deviceUsefulUtilization": 0.0133399209486166,
      "meanCycles": 3044.265625,
      "p50Cycles": 3378,
      "p95Cycles": 3378,
      "maximumCycles": 3378,
      "firstOffset": 212028,
      "lastOffset": 227316
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r0_p64_c32_p32_b64x16",
      "dimensions": {},
      "sampleCount": 64,
      "tileCount": 64,
      "phaseCount": 1,
      "phaseCycles": 11808,
      "phaseMs": 0.007872,
      "workCycles": 651264,
      "averageActiveTiles": 55.15447154471545,
      "usefulCycles": 25600.0,
      "estimatedWorkCycles": 651264,
      "usefulUtilization": 0.03930817610062893,
      "workLaneOccupancy": 0.390625,
      "deviceUsefulUtilization": 0.0014728408153646753,
      "meanCycles": 10176.0,
      "p50Cycles": 10176,
      "p95Cycles": 10176,
      "maximumCycles": 10176,
      "firstOffset": 210594,
      "lastOffset": 222402
    },
    {
      "name": "copy_strided_u64",
      "dimensions": {},
      "sampleCount": 2800,
      "tileCount": 1472,
      "phaseCount": 37,
      "phaseCycles": 11070,
      "phaseMs": 0.00738,
      "workCycles": 2348160,
      "averageActiveTiles": 212.11924119241192,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 838.6285714285714,
      "p50Cycles": 480,
      "p95Cycles": 1908,
      "maximumCycles": 1908,
      "firstOffset": 212094,
      "lastOffset": 359946
    },
    {
      "name": "attention_merge_v72_p80_k768_out16_p8_f16",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 25,
      "phaseCycles": 9084,
      "phaseMs": 0.006056,
      "workCycles": 6438528,
      "averageActiveTiles": 708.7767503302509,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 4374.0,
      "p50Cycles": 4374,
      "p95Cycles": 4374,
      "maximumCycles": 4374,
      "firstOffset": 355236,
      "lastOffset": 364320
    },
    {
      "name": "reduce_sum_f16",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 21,
      "phaseCycles": 5832,
      "phaseMs": 0.003888,
      "workCycles": 5414976,
      "averageActiveTiles": 928.4938271604939,
      "usefulCycles": 2129920.0,
      "estimatedWorkCycles": 5414976,
      "usefulUtilization": 0.393338770107199,
      "workLaneOccupancy": 0.9130315500685872,
      "deviceUsefulUtilization": 0.24810639947515953,
      "meanCycles": 3678.6521739130435,
      "p50Cycles": 3690,
      "p95Cycles": 3876,
      "maximumCycles": 3876,
      "firstOffset": 350256,
      "lastOffset": 356088
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r0_p64_c16_p16_b64x16",
      "dimensions": {},
      "sampleCount": 32,
      "tileCount": 32,
      "phaseCount": 1,
      "phaseCycles": 4176,
      "phaseMs": 0.002784,
      "workCycles": 62976,
      "averageActiveTiles": 15.080459770114942,
      "usefulCycles": 6400.0,
      "estimatedWorkCycles": 62976,
      "usefulUtilization": 0.1016260162601626,
      "workLaneOccupancy": 0.390625,
      "deviceUsefulUtilization": 0.0010411460936198568,
      "meanCycles": 1968.0,
      "p50Cycles": 1968,
      "p95Cycles": 1968,
      "maximumCycles": 1968,
      "firstOffset": 210060,
      "lastOffset": 214236
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r64_p64_c16_p16_b64x16",
      "dimensions": {},
      "sampleCount": 352,
      "tileCount": 352,
      "phaseCount": 3,
      "phaseCycles": 3960,
      "phaseMs": 0.00264,
      "workCycles": 857472,
      "averageActiveTiles": 216.53333333333333,
      "usefulCycles": 180224.0,
      "estimatedWorkCycles": 857472,
      "usefulUtilization": 0.21018062397372742,
      "workLaneOccupancy": 1.0,
      "deviceUsefulUtilization": 0.030917874396135265,
      "meanCycles": 2436.0,
      "p50Cycles": 2436,
      "p95Cycles": 2436,
      "maximumCycles": 2436,
      "firstOffset": 211020,
      "lastOffset": 214980
    },
    {
      "name": "static_copy_u32",
      "dimensions": {},
      "sampleCount": 16,
      "tileCount": 16,
      "phaseCount": 1,
      "phaseCycles": 294,
      "phaseMs": 0.000196,
      "workCycles": 3264,
      "averageActiveTiles": 11.10204081632653,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 204.0,
      "p50Cycles": 204,
      "p95Cycles": 204,
      "maximumCycles": 204,
      "firstOffset": 212508,
      "lastOffset": 212802
    }
  ],
  "samples": [
    {
      "physicalTile": 0,
      "offset": 273552,
      "duration": 48552,
      "phase": 36,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "505"
        },
        {
          "name": "kernelSpec",
          "value": "AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 }"
        },
        {
          "name": "outputElements",
          "value": "13312"
        },
        {
          "name": "input0Elements",
          "value": "11664"
        },
        {
          "name": "usefulCycles",
          "value": "22976"
        },
        {
          "name": "physicalWorkCycles",
          "value": "23040"
        },
        {
          "name": "workEstimateBasis",
          "value": "softmax: arithmetic/conversion issue slots, excluding loads and row addressing"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[16, 729, 0]"
        }
      ]
    },
    {
      "physicalTile": 1,
      "offset": 273174,
      "duration": 48552,
      "phase": 33,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "505"
        },
        {
          "name": "kernelSpec",
          "value": "AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 }"
        },
        {
          "name": "outputElements",
          "value": "13312"
        },
        {
          "name": "input0Elements",
          "value": "11664"
        },
        {
          "name": "usefulCycles",
          "value": "22976"
        },
        {
          "name": "physicalWorkCycles",
          "value": "23040"
        },
        {
          "name": "workEstimateBasis",
          "value": "softmax: arithmetic/conversion issue slots, excluding loads and row addressing"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[16, 729, 0]"
        }
      ]
    },
    {
      "physicalTile": 2,
      "offset": 273552,
      "duration": 48552,
      "phase": 36,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "505"
        },
        {
          "name": "kernelSpec",
          "value": "AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 }"
        },
        {
          "name": "outputElements",
          "value": "13312"
        },
        {
          "name": "input0Elements",
          "value": "11664"
        },
        {
          "name": "usefulCycles",
          "value": "22976"
        },
        {
          "name": "physicalWorkCycles",
          "value": "23040"
        },
        {
          "name": "workEstimateBasis",
          "value": "softmax: arithmetic/conversion issue slots, excluding loads and row addressing"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[16, 729, 0]"
        }
      ]
    }
  ]
}
