{
  "clockHz": 1500000000,
  "tileCount": 1472,
  "sampleCount": 187394,
  "matchedSampleCount": 36947,
  "profileSpanCycles": 10131456,
  "profileSpanMs": 6.754304,
  "atOffset": null,
  "groups": [
    {
      "name": "exchange",
      "dimensions": {},
      "sampleCount": 7360,
      "tileCount": 1472,
      "phaseCount": 5,
      "phaseCycles": 65934,
      "phaseMs": 0.043956,
      "workCycles": 53564220,
      "averageActiveTiles": 812.3914823914824,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 7277.747282608696,
      "p50Cycles": 6996,
      "p95Cycles": 12852,
      "maximumCycles": 22818,
      "firstOffset": 106380,
      "lastOffset": 196656
    },
    {
      "name": "attention_softmax_d72_p768_tail_f8_sm4",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 20,
      "phaseCycles": 28590,
      "phaseMs": 0.01906,
      "workCycles": 38108064,
      "averageActiveTiles": 1332.9158447009445,
      "usefulCycles": 16866144.0,
      "estimatedWorkCycles": 38108064,
      "usefulUtilization": 0.4425872697180313,
      "workLaneOccupancy": 0.9972413793103448,
      "deviceUsefulUtilization": 0.4007687394497924,
      "meanCycles": 25888.630434782608,
      "p50Cycles": 25890,
      "p95Cycles": 25890,
      "maximumCycles": 25890,
      "firstOffset": 148254,
      "lastOffset": 176844
    },
    {
      "name": "fill_zero_u64",
      "dimensions": {},
      "sampleCount": 14176,
      "tileCount": 1472,
      "phaseCount": 52,
      "phaseCycles": 21240,
      "phaseMs": 0.01416,
      "workCycles": 4881888,
      "averageActiveTiles": 229.84406779661018,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 2760192,
      "usefulUtilization": 0.0,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": 0.0,
      "meanCycles": 344.37697516930024,
      "p50Cycles": 306,
      "p95Cycles": 564,
      "maximumCycles": 564,
      "firstOffset": 106110,
      "lastOffset": 201732
    },
    {
      "name": "copy_u64",
      "dimensions": {},
      "sampleCount": 4443,
      "tileCount": 1472,
      "phaseCount": 48,
      "phaseCycles": 13368,
      "phaseMs": 0.008912,
      "workCycles": 3684534,
      "averageActiveTiles": 275.62342908438063,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 829.2896691424713,
      "p50Cycles": 960,
      "p95Cycles": 1236,
      "maximumCycles": 1236,
      "firstOffset": 107700,
      "lastOffset": 202248
    },
    {
      "name": "gemm_f16_init_small_rows_interleaved_k80_c80_r81_r81",
      "dimensions": {},
      "sampleCount": 864,
      "tileCount": 864,
      "phaseCount": 12,
      "phaseCycles": 12684,
      "phaseMs": 0.008456,
      "workCycles": 10715328,
      "averageActiveTiles": 844.7909176915799,
      "usefulCycles": 6298560.0,
      "estimatedWorkCycles": 10715328,
      "usefulUtilization": 0.5878084179970973,
      "workLaneOccupancy": 0.9,
      "deviceUsefulUtilization": 0.33734729134959485,
      "meanCycles": 12402.0,
      "p50Cycles": 12402,
      "p95Cycles": 12402,
      "maximumCycles": 12402,
      "firstOffset": 129504,
      "lastOffset": 142188
    },
    {
      "name": "gemm_f16_init_small_rows_interleaved_k80_c64_r81_r81",
      "dimensions": {},
      "sampleCount": 576,
      "tileCount": 576,
      "phaseCount": 13,
      "phaseCycles": 10584,
      "phaseMs": 0.007056,
      "workCycles": 5761152,
      "averageActiveTiles": 544.3265306122449,
      "usefulCycles": 3267378.0,
      "estimatedWorkCycles": 5761152,
      "usefulUtilization": 0.5671396970605879,
      "workLaneOccupancy": 0.875390625,
      "deviceUsefulUtilization": 0.2097209128216504,
      "meanCycles": 10002.0,
      "p50Cycles": 10002,
      "p95Cycles": 10002,
      "maximumCycles": 10002,
      "firstOffset": 128916,
      "lastOffset": 139500
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k128_c80_r56_r57",
      "dimensions": {},
      "sampleCount": 1152,
      "tileCount": 1152,
      "phaseCount": 20,
      "phaseCycles": 9654,
      "phaseMs": 0.006436,
      "workCycles": 9303552,
      "averageActiveTiles": 963.6991920447483,
      "usefulCycles": 4408992.0,
      "estimatedWorkCycles": 9303552,
      "usefulUtilization": 0.4739041604754829,
      "workLaneOccupancy": 0.854296875,
      "deviceUsefulUtilization": 0.3102588699435242,
      "meanCycles": 8076.0,
      "p50Cycles": 8076,
      "p95Cycles": 8076,
      "maximumCycles": 8076,
      "firstOffset": 184038,
      "lastOffset": 193692
    },
    {
      "name": "gemm_f8_init_large_rows_interleaved_k128_c80_r56_r57",
      "dimensions": {},
      "sampleCount": 96,
      "tileCount": 96,
      "phaseCount": 6,
      "phaseCycles": 8658,
      "phaseMs": 0.005772,
      "workCycles": 779904,
      "averageActiveTiles": 90.07900207900208,
      "usefulCycles": 373977.0,
      "estimatedWorkCycles": 779904,
      "usefulUtilization": 0.47951670974889216,
      "workLaneOccupancy": 0.854296875,
      "deviceUsefulUtilization": 0.02934401269999096,
      "meanCycles": 8124.0,
      "p50Cycles": 8124,
      "p95Cycles": 8124,
      "maximumCycles": 8124,
      "firstOffset": 185088,
      "lastOffset": 193746
    },
    {
      "name": "attention_merge_v72_p80_k768_out16_p8_f16",
      "dimensions": {},
      "sampleCount": 1472,
      "tileCount": 1472,
      "phaseCount": 33,
      "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": 198678,
      "lastOffset": 205656
    },
    {
      "name": "copy_strided_u64",
      "dimensions": {},
      "sampleCount": 1720,
      "tileCount": 1472,
      "phaseCount": 41,
      "phaseCycles": 5496,
      "phaseMs": 0.003664,
      "workCycles": 709056,
      "averageActiveTiles": 129.0131004366812,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 412.2418604651163,
      "p50Cycles": 420,
      "p95Cycles": 420,
      "maximumCycles": 480,
      "firstOffset": 115134,
      "lastOffset": 202668
    },
    {
      "name": "cast_f16_f8",
      "dimensions": {},
      "sampleCount": 960,
      "tileCount": 960,
      "phaseCount": 13,
      "phaseCycles": 5220,
      "phaseMs": 0.00348,
      "workCycles": 2286720,
      "averageActiveTiles": 438.0689655172414,
      "usefulCycles": 116640.0,
      "estimatedWorkCycles": 2286720,
      "usefulUtilization": 0.05100755667506297,
      "workLaneOccupancy": 0.94921875,
      "deviceUsefulUtilization": 0.015179910044977511,
      "meanCycles": 2382.0,
      "p50Cycles": 2382,
      "p95Cycles": 2382,
      "maximumCycles": 2382,
      "firstOffset": 115758,
      "lastOffset": 120978
    },
    {
      "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": 195762,
      "lastOffset": 199902
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r64_p64_c16_p16_b64x16",
      "dimensions": {},
      "sampleCount": 880,
      "tileCount": 880,
      "phaseCount": 13,
      "phaseCycles": 3924,
      "phaseMs": 0.002616,
      "workCycles": 2143680,
      "averageActiveTiles": 546.2996941896024,
      "usefulCycles": 450560.0,
      "estimatedWorkCycles": 2143680,
      "usefulUtilization": 0.21018062397372742,
      "workLaneOccupancy": 1.0,
      "deviceUsefulUtilization": 0.07800381154988255,
      "meanCycles": 2436.0,
      "p50Cycles": 2436,
      "p95Cycles": 2436,
      "maximumCycles": 2436,
      "firstOffset": 114672,
      "lastOffset": 118596
    },
    {
      "name": "rearrange_row_major_to_amp_f16_o2_r0_p64_c16_p16_b64x16",
      "dimensions": {},
      "sampleCount": 80,
      "tileCount": 80,
      "phaseCount": 7,
      "phaseCycles": 3798,
      "phaseMs": 0.002532,
      "workCycles": 157440,
      "averageActiveTiles": 41.45339652448657,
      "usefulCycles": 16000.0,
      "estimatedWorkCycles": 157440,
      "usefulUtilization": 0.1016260162601626,
      "workLaneOccupancy": 0.390625,
      "deviceUsefulUtilization": 0.002861918172035625,
      "meanCycles": 1968.0,
      "p50Cycles": 1968,
      "p95Cycles": 1968,
      "maximumCycles": 1968,
      "firstOffset": 113790,
      "lastOffset": 117588
    },
    {
      "name": "static_copy_u32",
      "dimensions": {},
      "sampleCount": 224,
      "tileCount": 184,
      "phaseCount": 8,
      "phaseCycles": 3408,
      "phaseMs": 0.002272,
      "workCycles": 63696,
      "averageActiveTiles": 18.690140845070424,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 284.35714285714283,
      "p50Cycles": 294,
      "p95Cycles": 294,
      "maximumCycles": 294,
      "firstOffset": 108570,
      "lastOffset": 116160
    }
  ],
  "samples": [
    {
      "physicalTile": 0,
      "offset": 149088,
      "duration": 25890,
      "phase": 41,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "513"
        },
        {
          "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": 149484,
      "duration": 25890,
      "phase": 34,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.12",
      "kernel": "attention_softmax_d72_p768_tail_f8_sm4",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "513"
        },
        {
          "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": 149094,
      "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": "513"
        },
        {
          "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]"
        }
      ]
    }
  ]
}
