{
  "clockHz": 1500000000,
  "tileCount": 1472,
  "sampleCount": 227478,
  "matchedSampleCount": 32594,
  "profileSpanCycles": 16600566,
  "profileSpanMs": 11.067044,
  "atOffset": null,
  "groups": [
    {
      "name": "exchange",
      "dimensions": {},
      "sampleCount": 7040,
      "tileCount": 1472,
      "phaseCount": 5,
      "phaseCycles": 94290,
      "phaseMs": 0.06286,
      "workCycles": 67886916,
      "averageActiveTiles": 719.9800190900413,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 9643.027840909092,
      "p50Cycles": 8964,
      "p95Cycles": 21408,
      "maximumCycles": 28860,
      "firstOffset": 206118,
      "lastOffset": 345516
    },
    {
      "name": "attention_softmax_d72_p768_tail_f8_sm4",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 16,
      "phaseCycles": 53334,
      "phaseMs": 0.035556,
      "workCycles": 71468520,
      "averageActiveTiles": 1340.0179997750029,
      "usefulCycles": 33499008.0,
      "estimatedWorkCycles": 71468520,
      "usefulUtilization": 0.4687239640613797,
      "workLaneOccupancy": 0.9972222222222222,
      "deviceUsefulUtilization": 0.42669738367400845,
      "meanCycles": 48551.983695652176,
      "p50Cycles": 48552,
      "p95Cycles": 48552,
      "maximumCycles": 48552,
      "firstOffset": 264954,
      "lastOffset": 318288
    },
    {
      "name": "cast_f16_f8",
      "dimensions": {},
      "sampleCount": 3353,
      "tileCount": 1472,
      "phaseCount": 16,
      "phaseCycles": 23154,
      "phaseMs": 0.015436,
      "workCycles": 10374456,
      "averageActiveTiles": 448.0632288157554,
      "usefulCycles": 653184.0,
      "estimatedWorkCycles": 10374456,
      "usefulUtilization": 0.06296079524555312,
      "workLaneOccupancy": 0.8107721639656816,
      "deviceUsefulUtilization": 0.019164685602262355,
      "meanCycles": 3094.081717864599,
      "p50Cycles": 2832,
      "p95Cycles": 3702,
      "maximumCycles": 3702,
      "firstOffset": 210006,
      "lastOffset": 235320
    },
    {
      "name": "fill_zero_u64",
      "dimensions": {},
      "sampleCount": 6208,
      "tileCount": 1472,
      "phaseCount": 40,
      "phaseCycles": 22920,
      "phaseMs": 0.01528,
      "workCycles": 2865024,
      "averageActiveTiles": 125.00104712041885,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 524160,
      "usefulUtilization": 0.0,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": 0.0,
      "meanCycles": 461.50515463917526,
      "p50Cycles": 396,
      "p95Cycles": 732,
      "maximumCycles": 732,
      "firstOffset": 206118,
      "lastOffset": 353292
    },
    {
      "name": "copy_u64",
      "dimensions": {},
      "sampleCount": 4697,
      "tileCount": 1472,
      "phaseCount": 37,
      "phaseCycles": 16206,
      "phaseMs": 0.010804,
      "workCycles": 6376686,
      "averageActiveTiles": 393.4768604220659,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 1357.608260591867,
      "p50Cycles": 1248,
      "p95Cycles": 2718,
      "maximumCycles": 2718,
      "firstOffset": 207288,
      "lastOffset": 354066
    },
    {
      "name": "gemm_f8_init_large_rows_interleaved_k96_c112_r121_r122",
      "dimensions": {},
      "sampleCount": 384,
      "tileCount": 384,
      "phaseCount": 4,
      "phaseCycles": 14196,
      "phaseMs": 0.009464,
      "workCycles": 5398272,
      "averageActiveTiles": 380.2671174978867,
      "usefulCycles": 2951424.0,
      "estimatedWorkCycles": 5398272,
      "usefulUtilization": 0.5467349551856594,
      "workLaneOccupancy": 0.75,
      "deviceUsefulUtilization": 0.1412400308721379,
      "meanCycles": 14058.0,
      "p50Cycles": 14058,
      "p95Cycles": 14058,
      "maximumCycles": 14058,
      "firstOffset": 243132,
      "lastOffset": 257328
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k96_c112_r121_r122",
      "dimensions": {},
      "sampleCount": 384,
      "tileCount": 384,
      "phaseCount": 4,
      "phaseCycles": 14166,
      "phaseMs": 0.009444,
      "workCycles": 5384448,
      "averageActiveTiles": 380.09656925031766,
      "usefulCycles": 2927232.0,
      "estimatedWorkCycles": 5384448,
      "usefulUtilization": 0.5436456996148908,
      "workLaneOccupancy": 0.75,
      "deviceUsefulUtilization": 0.14037898458648693,
      "meanCycles": 14022.0,
      "p50Cycles": 14022,
      "p95Cycles": 14022,
      "maximumCycles": 14022,
      "firstOffset": 243156,
      "lastOffset": 257322
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k128_c80_r104_r105",
      "dimensions": {},
      "sampleCount": 1152,
      "tileCount": 1152,
      "phaseCount": 14,
      "phaseCycles": 14040,
      "phaseMs": 0.00936,
      "workCycles": 13727232,
      "averageActiveTiles": 977.723076923077,
      "usefulCycles": 8188128.0,
      "estimatedWorkCycles": 13727232,
      "usefulUtilization": 0.596487915407855,
      "workLaneOccupancy": 0.854296875,
      "deviceUsefulUtilization": 0.39619565217391306,
      "meanCycles": 11916.0,
      "p50Cycles": 11916,
      "p95Cycles": 11916,
      "maximumCycles": 11916,
      "firstOffset": 326004,
      "lastOffset": 340044
    },
    {
      "name": "gemm_f8_init_large_rows_interleaved_k128_c80_r104_r105",
      "dimensions": {},
      "sampleCount": 192,
      "tileCount": 192,
      "phaseCount": 4,
      "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": 326808,
      "lastOffset": 340386
    },
    {
      "name": "gemm_f8_init_large_rows_interleaved_k96_c96_r121_r122",
      "dimensions": {},
      "sampleCount": 288,
      "tileCount": 288,
      "phaseCount": 3,
      "phaseCycles": 12678,
      "phaseMs": 0.008452,
      "workCycles": 3487104,
      "averageActiveTiles": 275.0515854235684,
      "usefulCycles": 1851228.0,
      "estimatedWorkCycles": 3487104,
      "usefulUtilization": 0.5308783448959365,
      "workLaneOccupancy": 0.7317708333333334,
      "deviceUsefulUtilization": 0.09919764295561637,
      "meanCycles": 12108.0,
      "p50Cycles": 12108,
      "p95Cycles": 12108,
      "maximumCycles": 12108,
      "firstOffset": 242556,
      "lastOffset": 255234
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k96_c96_r121_r122",
      "dimensions": {},
      "sampleCount": 288,
      "tileCount": 288,
      "phaseCount": 5,
      "phaseCycles": 12318,
      "phaseMs": 0.008212,
      "workCycles": 3478464,
      "averageActiveTiles": 282.38869946419874,
      "usefulCycles": 1836054.0,
      "estimatedWorkCycles": 3478464,
      "usefulUtilization": 0.5278346994535519,
      "workLaneOccupancy": 0.7317708333333334,
      "deviceUsefulUtilization": 0.10125988743937822,
      "meanCycles": 12078.0,
      "p50Cycles": 12078,
      "p95Cycles": 12078,
      "maximumCycles": 12078,
      "firstOffset": 242580,
      "lastOffset": 254898
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r64_p64_c32_p32_b64x16",
      "dimensions": {},
      "sampleCount": 704,
      "tileCount": 704,
      "phaseCount": 5,
      "phaseCycles": 12042,
      "phaseMs": 0.008028,
      "workCycles": 7569408,
      "averageActiveTiles": 628.5839561534629,
      "usefulCycles": 720896.0,
      "estimatedWorkCycles": 7569408,
      "usefulUtilization": 0.09523809523809523,
      "workLaneOccupancy": 1.0,
      "deviceUsefulUtilization": 0.04066925182152322,
      "meanCycles": 10752.0,
      "p50Cycles": 10752,
      "p95Cycles": 10752,
      "maximumCycles": 10752,
      "firstOffset": 217068,
      "lastOffset": 229110
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r0_p64_c32_p32_b64x16",
      "dimensions": {},
      "sampleCount": 64,
      "tileCount": 64,
      "phaseCount": 1,
      "phaseCycles": 11538,
      "phaseMs": 0.007692,
      "workCycles": 651264,
      "averageActiveTiles": 56.44513780551222,
      "usefulCycles": 25600.0,
      "estimatedWorkCycles": 651264,
      "usefulUtilization": 0.03930817610062893,
      "workLaneOccupancy": 0.390625,
      "deviceUsefulUtilization": 0.0015073066690783573,
      "meanCycles": 10176.0,
      "p50Cycles": 10176,
      "p95Cycles": 10176,
      "maximumCycles": 10176,
      "firstOffset": 216420,
      "lastOffset": 227958
    },
    {
      "name": "copy_strided_u64",
      "dimensions": {},
      "sampleCount": 2704,
      "tileCount": 1472,
      "phaseCount": 29,
      "phaseCycles": 10848,
      "phaseMs": 0.007232,
      "workCycles": 2311296,
      "averageActiveTiles": 213.06194690265488,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 854.7692307692307,
      "p50Cycles": 480,
      "p95Cycles": 1908,
      "maximumCycles": 1908,
      "firstOffset": 207654,
      "lastOffset": 354546
    },
    {
      "name": "sync",
      "dimensions": {},
      "sampleCount": 320,
      "tileCount": 320,
      "phaseCount": 1,
      "phaseCycles": 9144,
      "phaseMs": 0.006096,
      "workCycles": 2611242,
      "averageActiveTiles": 285.56889763779526,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 8160.13125,
      "p50Cycles": 8328,
      "p95Cycles": 9102,
      "maximumCycles": 9138,
      "firstOffset": 206106,
      "lastOffset": 215250
    },
    {
      "name": "attention_merge_v72_p80_k768_out16_p8_f16",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 23,
      "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": 349836,
      "lastOffset": 358920
    },
    {
      "name": "reduce_sum_f16",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 18,
      "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": 344856,
      "lastOffset": 350688
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r64_p64_c16_p16_b64x16",
      "dimensions": {},
      "sampleCount": 352,
      "tileCount": 352,
      "phaseCount": 3,
      "phaseCycles": 3264,
      "phaseMs": 0.002176,
      "workCycles": 857472,
      "averageActiveTiles": 262.70588235294116,
      "usefulCycles": 180224.0,
      "estimatedWorkCycles": 857472,
      "usefulUtilization": 0.21018062397372742,
      "workLaneOccupancy": 1.0,
      "deviceUsefulUtilization": 0.03751065643648764,
      "meanCycles": 2436.0,
      "p50Cycles": 2436,
      "p95Cycles": 2436,
      "maximumCycles": 2436,
      "firstOffset": 217602,
      "lastOffset": 220866
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r0_p64_c16_p16_b64x16",
      "dimensions": {},
      "sampleCount": 32,
      "tileCount": 32,
      "phaseCount": 1,
      "phaseCycles": 2682,
      "phaseMs": 0.001788,
      "workCycles": 62976,
      "averageActiveTiles": 23.480984340044742,
      "usefulCycles": 6400.0,
      "estimatedWorkCycles": 62976,
      "usefulUtilization": 0.1016260162601626,
      "workLaneOccupancy": 0.390625,
      "deviceUsefulUtilization": 0.0016211133806698442,
      "meanCycles": 1968.0,
      "p50Cycles": 1968,
      "p95Cycles": 1968,
      "maximumCycles": 1968,
      "firstOffset": 217260,
      "lastOffset": 219942
    },
    {
      "name": "static_copy_u32",
      "dimensions": {},
      "sampleCount": 16,
      "tileCount": 16,
      "phaseCount": 1,
      "phaseCycles": 264,
      "phaseMs": 0.000176,
      "workCycles": 3264,
      "averageActiveTiles": 12.363636363636363,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 204.0,
      "p50Cycles": 204,
      "p95Cycles": 204,
      "maximumCycles": 204,
      "firstOffset": 208038,
      "lastOffset": 208302
    }
  ],
  "samples": [
    {
      "physicalTile": 0,
      "offset": 267642,
      "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": "507"
        },
        {
          "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": 267060,
      "duration": 48552,
      "phase": 30,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "507"
        },
        {
          "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": 267642,
      "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": "507"
        },
        {
          "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]"
        }
      ]
    }
  ]
}
