{
  "profile": "../poprt/profile_logs_f8weight/inference/profile.pop",
  "kernels": [
    {
      "step": 101,
      "kind": "OnTileExecute",
      "stage": "fused QKV projection",
      "name": "/visual/trunk/blocks/blocks.0/attn/qkv/MatMul/1570/matmulGrouped/Conv_1/Convolve",
      "cycles": 37277,
      "active_tiles": 1458,
      "start": 597717142,
      "end": 597755322,
      "max_exchange_code_bytes": 0,
      "vertices": "poplin::ConvPartial1x1Out<half,half,true,false,16,4,false>"
    },
    {
      "step": 121,
      "kind": "OnTileExecute",
      "stage": "QK transpose",
      "name": "/visual/trunk/blocks/blocks.0/attn/MatMul/1583/matmulGrouped/Conv_1/Convolve",
      "cycles": 8488,
      "active_tiles": 1472,
      "start": 597782782,
      "end": 597791385,
      "max_exchange_code_bytes": 0,
      "vertices": "poplin::ConvPartial1x1Out<half,half,true,false,16,4,false>"
    },
    {
      "step": 146,
      "kind": "OnTileExecute",
      "stage": "PV",
      "name": "/visual/trunk/blocks/blocks.0/attn/MatMul_1/1585/matmulGrouped/Conv_1/Convolve",
      "cycles": 12900,
      "active_tiles": 1440,
      "start": 597953551,
      "end": 597986460,
      "max_exchange_code_bytes": 0,
      "vertices": "poplin::ConvPartial1x1Out<half,half,true,false,16,4,false>"
    },
    {
      "step": 157,
      "kind": "OnTileExecute",
      "stage": "output projection",
      "name": "/visual/trunk/blocks/blocks.0/attn/proj/MatMul/1588/matmulGrouped/Conv_1/Convolve",
      "cycles": 14153,
      "active_tiles": 1440,
      "start": 597999519,
      "end": 598014540,
      "max_exchange_code_bytes": 0,
      "vertices": "poplin::ConvPartial1x1Out<half,half,true,false,16,4,false>"
    },
    {
      "step": 211,
      "kind": "OnTileExecute",
      "stage": "MLP up",
      "name": "/visual/trunk/blocks/blocks.0/mlp/fc1/MatMul/1601/matmulGrouped/Conv_1/Convolve",
      "cycles": 48248,
      "active_tiles": 1440,
      "start": 598060704,
      "end": 598110220,
      "max_exchange_code_bytes": 0,
      "vertices": "poplin::ConvPartial1x1Out<half,half,true,false,16,4,false>"
    },
    {
      "step": 240,
      "kind": "OnTileExecute",
      "stage": "MLP down",
      "name": "/visual/trunk/blocks/blocks.0/mlp/fc2/MatMul/1608/matmulGrouped/Conv_1/Convolve",
      "cycles": 52187,
      "active_tiles": 1440,
      "start": 598182986,
      "end": 598242933,
      "max_exchange_code_bytes": 0,
      "vertices": "poplin::ConvPartial1x1Out<half,half,true,false,16,4,false>"
    }
  ],
  "attention_exchanges": [
    {
      "name": "copy1_Copy_/visual/trunk/blocks/blocks.0/norm1/Sub/1561/Op/Subtract/Out_to_/visual/trunk/blocks/blocks.0/attn/qkv/MatMul/1570/matmulGrouped/Conv_1/actsRearranged/ExchangePre",
      "maximum": 64,
      "mean": 40.95108695652174
    },
    {
      "name": "cs23_/visual/trunk/blocks/blocks.0/attn/qkv/MatMul/1570/matmulGrouped/Conv_1/Convolve/ExchangePre",
      "maximum": 324,
      "mean": 171.89945652173913
    },
    {
      "name": "cs24_/visual/trunk/blocks/blocks.0/attn/qkv/MatMul/1570/matmulGrouped/Conv_1/Reduce0/ExchangePre",
      "maximum": 200,
      "mean": 157.14673913043478
    },
    {
      "name": "cs25_/visual/trunk/blocks/blocks.0/attn/qkv/Add/1773/Op/Add/ExchangePre",
      "maximum": 40,
      "mean": 17.630434782608695
    },
    {
      "name": "cs26_/visual/trunk/blocks/blocks.0/attn/Mul_1/1757/Op/Multiply/ExchangePre",
      "maximum": 16,
      "mean": 4.627717391304348
    },
    {
      "name": "cs27_/visual/trunk/blocks/blocks.0/attn/Mul/1755/Op/Multiply/ExchangePre",
      "maximum": 16,
      "mean": 4.619565217391305
    },
    {
      "name": "copy617_Copy_{/visual/trunk/blocks/blocks.0/attn/qkv/MatMul/1570/matmulGrouped/Conv_1/partialReduceOut,ValuePadder/padding}_to_{/visual/trunk/blocks/blocks.0/attn/MatMul/1583/matmulGrouped/Conv_1/actsRearranged,/visual/trunk/blocks/blocks.0/attn/MatMul/1583/matmulGrouped/Conv_1/weightsRearranged}/ExchangePre",
      "maximum": 1544,
      "mean": 481.3233695652174
    },
    {
      "name": "cs28_/visual/trunk/blocks/blocks.0/attn/MatMul/1583/matmulGrouped/Conv_1/Convolve/ExchangePre",
      "maximum": 248,
      "mean": 149.04347826086956
    },
    {
      "name": "copy5_Copy_{/visual/trunk/blocks/blocks.0/attn/MatMul/1583/matmulGrouped/Conv_1/partials,/visual/trunk/blocks/blocks.0/attn/Softmax/1584/alignPad}_to_/visual/trunk/blocks/blocks.0/attn/Softmax/1584/Out/ExchangePre",
      "maximum": 7372,
      "mean": 5242.347826086957
    },
    {
      "name": "copy592_Copy_{/visual/trunk/blocks/blocks.0/attn/qkv/MatMul/1570/matmulGrouped/Conv_1/partialReduceOut,ValuePadder/padding}_to_/visual/trunk/blocks/blocks.0/attn/MatMul_1/1585/matmulGrouped/Conv_1/preRegroup/ExchangePre",
      "maximum": 1300,
      "mean": 304.2771739130435
    },
    {
      "name": "copy7_Copy_/visual/trunk/blocks/blocks.0/attn/MatMul_1/1585/matmulGrouped/Conv_1/internal/partialTranspose_to_/visual/trunk/blocks/blocks.0/attn/MatMul_1/1585/matmulGrouped/Conv_1/weightsRearranged/ExchangePre",
      "maximum": 64,
      "mean": 35.85054347826087
    },
    {
      "name": "cs36_/visual/trunk/blocks/blocks.0/attn/MatMul_1/1585/matmulGrouped/Conv_1/Convolve/ExchangePre",
      "maximum": 2212,
      "mean": 1938.0190217391305
    },
    {
      "name": "cs38_/visual/trunk/blocks/blocks.0/attn/MatMul_1/1585/matmulGrouped/Conv_1/Reduce0/ExchangePre",
      "maximum": 208,
      "mean": 169.0054347826087
    },
    {
      "name": "copy8_Copy_/visual/trunk/blocks/blocks.0/attn/MatMul_1/1585/matmulGrouped/Conv_1/partialReduceOut_to_/visual/trunk/blocks/blocks.0/attn/proj/MatMul/1588/matmulGrouped/Conv_1/actsRearranged/ExchangePre",
      "maximum": 680,
      "mean": 351.9755434782609
    },
    {
      "name": "cs41_/visual/trunk/blocks/blocks.0/attn/proj/MatMul/1588/matmulGrouped/Conv_1/Convolve/ExchangePre",
      "maximum": 192,
      "mean": 119.61413043478261
    },
    {
      "name": "cs42_/visual/trunk/blocks/blocks.0/attn/proj/MatMul/1588/matmulGrouped/Conv_1/Reduce0/ExchangePre",
      "maximum": 268,
      "mean": 209.5108695652174
    },
    {
      "name": "cs43_/visual/trunk/blocks/blocks.0/attn/proj/Add/1751/Op/Add/ExchangePre",
      "maximum": 32,
      "mean": 15.755434782608695
    }
  ],
  "exchange_memory_per_tile": {
    "min(b)": 4436,
    "avg(b)": 15602.703804347826,
    "max(b)": 37092
  },
  "device_spans": [
    15987664,
    15987659
  ],
  "qkv_executions": 54,
  "allocated_memory": {
    "min(total)": 86179,
    "avg(total)": 550544.296875,
    "max(total)": 596494
  }
}