{
  "clockHz": 1500000000,
  "tileCount": 1472,
  "sampleCount": 187988,
  "matchedSampleCount": 6832,
  "profileSpanCycles": 10158156,
  "profileSpanMs": 6.772104,
  "atOffset": null,
  "groups": [
    {
      "name": "exchange",
      "dimensions": {},
      "sampleCount": 2916,
      "tileCount": 1458,
      "phaseCount": 2,
      "phaseCycles": 43182,
      "phaseMs": 0.028788,
      "workCycles": 45984150,
      "averageActiveTiles": 1064.891621508962,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 15769.598765432098,
      "p50Cycles": 14514,
      "p95Cycles": 27684,
      "maximumCycles": 28014,
      "firstOffset": 240672,
      "lastOffset": 313704
    },
    {
      "name": "gemm_f8_init_large_rows_interleaved_k192_c160_r80_r82",
      "dimensions": {},
      "sampleCount": 780,
      "tileCount": 780,
      "phaseCount": 29,
      "phaseCycles": 32730,
      "phaseMs": 0.02182,
      "workCycles": 23016240,
      "averageActiveTiles": 703.2153987167736,
      "usefulCycles": 15350400.0,
      "estimatedWorkCycles": 23016240,
      "usefulUtilization": 0.6669377795851973,
      "workLaneOccupancy": 1.0,
      "deviceUsefulUtilization": 0.31861475311839954,
      "meanCycles": 29508.0,
      "p50Cycles": 29508,
      "p95Cycles": 29508,
      "maximumCycles": 29508,
      "firstOffset": 265722,
      "lastOffset": 298452
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k192_c160_r80_r82",
      "dimensions": {},
      "sampleCount": 624,
      "tileCount": 624,
      "phaseCount": 22,
      "phaseCycles": 30048,
      "phaseMs": 0.020032,
      "workCycles": 18094752,
      "averageActiveTiles": 602.1948881789137,
      "usefulCycles": 11943360.0,
      "estimatedWorkCycles": 18094752,
      "usefulUtilization": 0.660045520380716,
      "workLaneOccupancy": 0.996875,
      "deviceUsefulUtilization": 0.2700244825670232,
      "meanCycles": 28998.0,
      "p50Cycles": 28998,
      "p95Cycles": 28998,
      "maximumCycles": 28998,
      "firstOffset": 267930,
      "lastOffset": 297978
    },
    {
      "name": "gemm_f8_init_large_rows_interleaved_k192_c144_r80_r82",
      "dimensions": {},
      "sampleCount": 30,
      "tileCount": 30,
      "phaseCount": 15,
      "phaseCycles": 29796,
      "phaseMs": 0.019864,
      "workCycles": 797940,
      "averageActiveTiles": 26.780104712041886,
      "usefulCycles": 531360.0,
      "estimatedWorkCycles": 797940,
      "usefulUtilization": 0.6659147304308595,
      "workLaneOccupancy": 1.0,
      "deviceUsefulUtilization": 0.01211499063194943,
      "meanCycles": 26598.0,
      "p50Cycles": 26598,
      "p95Cycles": 26598,
      "maximumCycles": 26598,
      "firstOffset": 265734,
      "lastOffset": 295530
    },
    {
      "name": "gemm_f8_init_small_rows_interleaved_k192_c144_r80_r82",
      "dimensions": {},
      "sampleCount": 24,
      "tileCount": 24,
      "phaseCount": 14,
      "phaseCycles": 27180,
      "phaseMs": 0.01812,
      "workCycles": 627408,
      "averageActiveTiles": 23.08344370860927,
      "usefulCycles": 413424.0,
      "estimatedWorkCycles": 627408,
      "usefulUtilization": 0.6589396373651595,
      "workLaneOccupancy": 0.996875,
      "deviceUsefulUtilization": 0.010333285344082925,
      "meanCycles": 26142.0,
      "p50Cycles": 26142,
      "p95Cycles": 26142,
      "maximumCycles": 26142,
      "firstOffset": 267930,
      "lastOffset": 295110
    },
    {
      "name": "reduce_bias_gelu_f16",
      "dimensions": {},
      "sampleCount": 1458,
      "tileCount": 1458,
      "phaseCount": 33,
      "phaseCycles": 26604,
      "phaseMs": 0.017736,
      "workCycles": 21570438,
      "averageActiveTiles": 810.7967974740641,
      "usefulCycles": 13334868.0,
      "estimatedWorkCycles": 21570438,
      "usefulUtilization": 0.6182010768626951,
      "workLaneOccupancy": 0.9986301369863013,
      "deviceUsefulUtilization": 0.34051321556745306,
      "meanCycles": 14794.539094650207,
      "p50Cycles": 17586,
      "p95Cycles": 17898,
      "maximumCycles": 17898,
      "firstOffset": 304998,
      "lastOffset": 331602
    },
    {
      "name": "sync",
      "dimensions": {},
      "sampleCount": 28,
      "tileCount": 14,
      "phaseCount": 2,
      "phaseCycles": 12864,
      "phaseMs": 0.008576,
      "workCycles": 179964,
      "averageActiveTiles": 13.989738805970148,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 0,
      "usefulUtilization": null,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": null,
      "meanCycles": 6427.285714285715,
      "p50Cycles": 492,
      "p95Cycles": 12366,
      "maximumCycles": 12372,
      "firstOffset": 239196,
      "lastOffset": 299208
    },
    {
      "name": "fill_zero_u64",
      "dimensions": {},
      "sampleCount": 972,
      "tileCount": 162,
      "phaseCount": 26,
      "phaseCycles": 2304,
      "phaseMs": 0.001536,
      "workCycles": 239112,
      "averageActiveTiles": 103.78125,
      "usefulCycles": 0.0,
      "estimatedWorkCycles": 239112,
      "usefulUtilization": 0.0,
      "workLaneOccupancy": null,
      "deviceUsefulUtilization": 0.0,
      "meanCycles": 246.0,
      "p50Cycles": 246,
      "p95Cycles": 246,
      "maximumCycles": 246,
      "firstOffset": 240672,
      "lastOffset": 242976
    }
  ],
  "samples": [
    {
      "physicalTile": 0,
      "offset": 265722,
      "duration": 29508,
      "phase": 77,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.18",
      "kernel": "gemm_f8_init_large_rows_interleaved_k192_c160_r80_r82",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "523"
        },
        {
          "name": "kernelSpec",
          "value": "Gemm { multiply: F8F143 { scale_exponent: -4 }, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 192, output_columns: 160 }"
        },
        {
          "name": "outputElements",
          "value": "13120"
        },
        {
          "name": "input0Elements",
          "value": "15744"
        },
        {
          "name": "input1Elements",
          "value": "30720"
        },
        {
          "name": "usefulCycles",
          "value": "19680"
        },
        {
          "name": "physicalWorkCycles",
          "value": "19680"
        },
        {
          "name": "workEstimateBasis",
          "value": "AMP FLOPs / IPU21 peak"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[4294967288]"
        }
      ]
    },
    {
      "physicalTile": 1,
      "offset": 265764,
      "duration": 29508,
      "phase": 67,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.18",
      "kernel": "gemm_f8_init_large_rows_interleaved_k192_c160_r80_r82",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "523"
        },
        {
          "name": "kernelSpec",
          "value": "Gemm { multiply: F8F143 { scale_exponent: -4 }, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 192, output_columns: 160 }"
        },
        {
          "name": "outputElements",
          "value": "13120"
        },
        {
          "name": "input0Elements",
          "value": "15744"
        },
        {
          "name": "input1Elements",
          "value": "30720"
        },
        {
          "name": "usefulCycles",
          "value": "19680"
        },
        {
          "name": "physicalWorkCycles",
          "value": "19680"
        },
        {
          "name": "workEstimateBasis",
          "value": "AMP FLOPs / IPU21 peak"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[4294967288]"
        }
      ]
    },
    {
      "physicalTile": 2,
      "offset": 265722,
      "duration": 29508,
      "phase": 70,
      "epoch": 1,
      "kind": "compute",
      "operation": "operation.18",
      "kernel": "gemm_f8_init_large_rows_interleaved_k192_c160_r80_r82",
      "metadata": [
        {
          "name": "reason",
          "value": "OperatorKernel"
        },
        {
          "name": "value",
          "value": "523"
        },
        {
          "name": "kernelSpec",
          "value": "Gemm { multiply: F8F143 { scale_exponent: -4 }, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 192, output_columns: 160 }"
        },
        {
          "name": "outputElements",
          "value": "13120"
        },
        {
          "name": "input0Elements",
          "value": "15744"
        },
        {
          "name": "input1Elements",
          "value": "30720"
        },
        {
          "name": "usefulCycles",
          "value": "19680"
        },
        {
          "name": "physicalWorkCycles",
          "value": "19680"
        },
        {
          "name": "workEstimateBasis",
          "value": "AMP FLOPs / IPU21 peak"
        },
        {
          "name": "invocations",
          "value": "1"
        },
        {
          "name": "uniformInvocations",
          "value": "true"
        },
        {
          "name": "arguments",
          "value": "[4294967288]"
        }
      ]
    }
  ]
}
