{
  "clockHz": 1500000000,
  "tileCount": 1472,
  "sampleCount": 192402,
  "matchedSampleCount": 37552,
  "profileSpanCycles": 10142796,
  "profileSpanMs": 6.761864,
  "atOffset": null,
  "groups": [
    {
      "name": "exchange",
      "dimensions": {},
      "sampleCount": 7346,
      "tileCount": 1472,
      "phaseCount": 5,
      "phaseCycles": 61146,
      "phaseMs": 0.040764,
      "workCycles": 47319840,
      "averageActiveTiles": 773.8828377980572,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 6441.579090661585,
      "p50Cycles": 6108,
      "p95Cycles": 11466,
      "maximumCycles": 23598,
      "firstOffset": 112254,
      "lastOffset": 197238
    },
    {
      "name": "attention_softmax_d72_p768_tail_f8_sm4",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 20,
      "phaseCycles": 29034,
      "phaseMs": 0.019356,
      "workCycles": 38108064,
      "averageActiveTiles": 1312.5323413928497,
      "usefulCycles": 16866144.0,
      "estimatedWorkCycles": 38108064,
      "usefulUtilization": 0.4425872697180313,
      "workLaneOccupancy": 0.9972413793103448,
      "deviceUsefulUtilization": 0.39464001725113884,
      "meanCycles": 25888.630434782608,
      "p50Cycles": 25890,
      "p95Cycles": 25890,
      "maximumCycles": 25890,
      "firstOffset": 148518,
      "lastOffset": 177552
    },
    {
      "name": "fill_zero_u64",
      "dimensions": {},
      "sampleCount": 12704,
      "tileCount": 1472,
      "phaseCount": 52,
      "phaseCycles": 17640,
      "phaseMs": 0.01176,
      "workCycles": 4881888,
      "averageActiveTiles": 276.75102040816324,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 2725632,
      "usefulUtilization": 0.0,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": 0.0,
      "meanCycles": 384.27959697732996,
      "p50Cycles": 366,
      "p95Cycles": 564,
      "maximumCycles": 564,
      "firstOffset": 112266,
      "lastOffset": 202308
    },
    {
      "name": "cast_f16_f8",
      "dimensions": {},
      "sampleCount": 3872,
      "tileCount": 1472,
      "phaseCount": 26,
      "phaseCycles": 14880,
      "phaseMs": 0.00992,
      "workCycles": 13195008,
      "averageActiveTiles": 886.7612903225806,
      "usefulCycles": 1166400.0,
      "estimatedWorkCycles": 13195008,
      "usefulUtilization": 0.08839706652697747,
      "workLaneOccupancy": 0.7600400350306519,
      "deviceUsefulUtilization": 0.05325210378681627,
      "meanCycles": 3407.801652892562,
      "p50Cycles": 2628,
      "p95Cycles": 5460,
      "maximumCycles": 5460,
      "firstOffset": 109626,
      "lastOffset": 135306
    },
    {
      "name": "copy_u64",
      "dimensions": {},
      "sampleCount": 3712,
      "tileCount": 1472,
      "phaseCount": 40,
      "phaseCycles": 10770,
      "phaseMs": 0.00718,
      "workCycles": 3285120,
      "averageActiveTiles": 305.025069637883,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 885.0,
      "p50Cycles": 1008,
      "p95Cycles": 1236,
      "maximumCycles": 1236,
      "firstOffset": 148578,
      "lastOffset": 202824
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k128_c80_r56_r57",
      "dimensions": {},
      "sampleCount": 1152,
      "tileCount": 1152,
      "phaseCount": 20,
      "phaseCycles": 9558,
      "phaseMs": 0.006372,
      "workCycles": 9303552,
      "averageActiveTiles": 973.3785310734463,
      "usefulCycles": 4408992.0,
      "estimatedWorkCycles": 9303552,
      "usefulUtilization": 0.4739041604754829,
      "workLaneOccupancy": 0.854296875,
      "deviceUsefulUtilization": 0.31337509211495945,
      "meanCycles": 8076.0,
      "p50Cycles": 8076,
      "p95Cycles": 8076,
      "maximumCycles": 8076,
      "firstOffset": 184692,
      "lastOffset": 194250
    },
    {
      "name": "gemm_f8_init_large_rows_interleaved_k128_c80_r56_r57",
      "dimensions": {},
      "sampleCount": 96,
      "tileCount": 96,
      "phaseCount": 7,
      "phaseCycles": 8598,
      "phaseMs": 0.005732,
      "workCycles": 779904,
      "averageActiveTiles": 90.70760642009769,
      "usefulCycles": 373977.0,
      "estimatedWorkCycles": 779904,
      "usefulUtilization": 0.47951670974889216,
      "workLaneOccupancy": 0.854296875,
      "deviceUsefulUtilization": 0.02954878599168664,
      "meanCycles": 8124.0,
      "p50Cycles": 8124,
      "p95Cycles": 8124,
      "maximumCycles": 8124,
      "firstOffset": 185736,
      "lastOffset": 194334
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k96_c80_r81_r81",
      "dimensions": {},
      "sampleCount": 864,
      "tileCount": 864,
      "phaseCount": 11,
      "phaseCycles": 7962,
      "phaseMs": 0.005308,
      "workCycles": 6630336,
      "averageActiveTiles": 832.7475508666164,
      "usefulCycles": 3149280.0,
      "estimatedWorkCycles": 6630336,
      "usefulUtilization": 0.4749804534792807,
      "workLaneOccupancy": 0.75,
      "deviceUsefulUtilization": 0.26870843026113167,
      "meanCycles": 7674.0,
      "p50Cycles": 7674,
      "p95Cycles": 7674,
      "maximumCycles": 7674,
      "firstOffset": 135018,
      "lastOffset": 142980
    },
    {
      "name": "attention_merge_v72_p80_k768_out16_p8_f16",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 34,
      "phaseCycles": 6978,
      "phaseMs": 0.004652,
      "workCycles": 4398336,
      "averageActiveTiles": 630.3147033533963,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 2988.0,
      "p50Cycles": 2988,
      "p95Cycles": 2988,
      "maximumCycles": 2988,
      "firstOffset": 199254,
      "lastOffset": 206232
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k96_c64_r81_r81",
      "dimensions": {},
      "sampleCount": 576,
      "tileCount": 576,
      "phaseCount": 13,
      "phaseCycles": 6810,
      "phaseMs": 0.00454,
      "workCycles": 3583872,
      "averageActiveTiles": 526.2660792951542,
      "usefulCycles": 1633689.0,
      "estimatedWorkCycles": 3583872,
      "usefulUtilization": 0.45584468418514945,
      "workLaneOccupancy": 0.7294921875,
      "deviceUsefulUtilization": 0.16297255075656006,
      "meanCycles": 6222.0,
      "p50Cycles": 6222,
      "p95Cycles": 6222,
      "maximumCycles": 6222,
      "firstOffset": 133002,
      "lastOffset": 139812
    },
    {
      "name": "copy_strided_u64",
      "dimensions": {},
      "sampleCount": 1720,
      "tileCount": 1472,
      "phaseCount": 39,
      "phaseCycles": 5364,
      "phaseMs": 0.003576,
      "workCycles": 709056,
      "averageActiveTiles": 132.18791946308724,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 412.2418604651163,
      "p50Cycles": 420,
      "p95Cycles": 420,
      "maximumCycles": 480,
      "firstOffset": 110310,
      "lastOffset": 203244
    },
    {
      "name": "reduce_sum_f16",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 27,
      "phaseCycles": 4140,
      "phaseMs": 0.00276,
      "workCycles": 3028032,
      "averageActiveTiles": 731.408695652174,
      "usefulCycles": 1065120.0,
      "estimatedWorkCycles": 3028032,
      "usefulUtilization": 0.3517532179316467,
      "workLaneOccupancy": 0.9131687242798354,
      "deviceUsefulUtilization": 0.17477945809703843,
      "meanCycles": 2057.086956521739,
      "p50Cycles": 2052,
      "p95Cycles": 2238,
      "maximumCycles": 2238,
      "firstOffset": 196338,
      "lastOffset": 200478
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r64_p64_c16_p16_b64x16",
      "dimensions": {},
      "sampleCount": 880,
      "tileCount": 880,
      "phaseCount": 14,
      "phaseCycles": 3900,
      "phaseMs": 0.0026,
      "workCycles": 2143680,
      "averageActiveTiles": 549.6615384615385,
      "usefulCycles": 450560.0,
      "estimatedWorkCycles": 2143680,
      "usefulUtilization": 0.21018062397372742,
      "workLaneOccupancy": 1.0,
      "deviceUsefulUtilization": 0.07848383500557414,
      "meanCycles": 2436.0,
      "p50Cycles": 2436,
      "p95Cycles": 2436,
      "maximumCycles": 2436,
      "firstOffset": 116070,
      "lastOffset": 119970
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r0_p64_c16_p16_b64x16",
      "dimensions": {},
      "sampleCount": 80,
      "tileCount": 80,
      "phaseCount": 7,
      "phaseCycles": 3258,
      "phaseMs": 0.002172,
      "workCycles": 157440,
      "averageActiveTiles": 48.324125230202576,
      "usefulCycles": 16000.0,
      "estimatedWorkCycles": 157440,
      "usefulUtilization": 0.1016260162601626,
      "workLaneOccupancy": 0.390625,
      "deviceUsefulUtilization": 0.003336269250273574,
      "meanCycles": 1968.0,
      "p50Cycles": 1968,
      "p95Cycles": 1968,
      "maximumCycles": 1968,
      "firstOffset": 115782,
      "lastOffset": 119040
    },
    {
      "name": "static_copy_u32",
      "dimensions": {},
      "sampleCount": 120,
      "tileCount": 60,
      "phaseCount": 8,
      "phaseCycles": 2562,
      "phaseMs": 0.001708,
      "workCycles": 31680,
      "averageActiveTiles": 12.365339578454332,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 264.0,
      "p50Cycles": 294,
      "p95Cycles": 294,
      "maximumCycles": 294,
      "firstOffset": 110694,
      "lastOffset": 117534
    },
    {
      "name": "sync",
      "dimensions": {},
      "sampleCount": 14,
      "tileCount": 14,
      "phaseCount": 1,
      "phaseCycles": 1950,
      "phaseMs": 0.0013,
      "workCycles": 24942,
      "averageActiveTiles": 12.79076923076923,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 1781.5714285714287,
      "p50Cycles": 1656,
      "p95Cycles": 1950,
      "maximumCycles": 1950,
      "firstOffset": 112632,
      "lastOffset": 114582
    }
  ],
  "samples": [
    {
      "physicalTile": 0,
      "offset": 150696,
      "duration": 25890,
      "phase": 44,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "515"
        },
        {
          "name": "kernelSpec",
          "value": "AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 }"
        },
        {
          "name": "outputElements",
          "value": "6656"
        },
        {
          "name": "input0Elements",
          "value": "5832"
        },
        {
          "name": "usefulCycles",
          "value": "11568"
        },
        {
          "name": "physicalWorkCycles",
          "value": "11600"
        },
        {
          "name": "workEstimateBasis",
          "value": "softmax: arithmetic/conversion issue slots, excluding loads and row addressing"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[8, 729, 1]"
        }
      ]
    },
    {
      "physicalTile": 1,
      "offset": 151104,
      "duration": 25890,
      "phase": 37,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "515"
        },
        {
          "name": "kernelSpec",
          "value": "AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 }"
        },
        {
          "name": "outputElements",
          "value": "6656"
        },
        {
          "name": "input0Elements",
          "value": "5832"
        },
        {
          "name": "usefulCycles",
          "value": "11568"
        },
        {
          "name": "physicalWorkCycles",
          "value": "11600"
        },
        {
          "name": "workEstimateBasis",
          "value": "softmax: arithmetic/conversion issue slots, excluding loads and row addressing"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[8, 729, 1]"
        }
      ]
    },
    {
      "physicalTile": 2,
      "offset": 150696,
      "duration": 25890,
      "phase": 40,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "515"
        },
        {
          "name": "kernelSpec",
          "value": "AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 }"
        },
        {
          "name": "outputElements",
          "value": "6656"
        },
        {
          "name": "input0Elements",
          "value": "5832"
        },
        {
          "name": "usefulCycles",
          "value": "11568"
        },
        {
          "name": "physicalWorkCycles",
          "value": "11600"
        },
        {
          "name": "workEstimateBasis",
          "value": "softmax: arithmetic/conversion issue slots, excluding loads and row addressing"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[8, 729, 1]"
        }
      ]
    }
  ]
}
