{
  "validation": {
    "case": {
      "kind": "case",
      "operation": "row",
      "rows": 4096,
      "cols": 128,
      "pattern": "balanced",
      "mode": "bench"
    },
    "output_checks": [
      {
        "file": "staged_stream_2.f32",
        "elements": 524288,
        "max_abs_error": 2.4124771269562473e-07,
        "max_row_sum_error": 2.4125696063137525e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "staged_graph_2.f32",
        "elements": 524288,
        "max_abs_error": 2.4124771269562473e-07,
        "max_row_sum_error": 2.4125696063137525e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "fused_stream_2.f32",
        "elements": 524288,
        "max_abs_error": 2.4124771269562473e-07,
        "max_row_sum_error": 2.4125696063137525e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "fused_graph_2.f32",
        "elements": 524288,
        "max_abs_error": 2.4124771269562473e-07,
        "max_row_sum_error": 2.4125696063137525e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "cudnn_stream_2.f32",
        "elements": 524288,
        "max_abs_error": 2.539449394101112e-07,
        "max_row_sum_error": 2.5395899383440224e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "cudnn_graph_2.f32",
        "elements": 524288,
        "max_abs_error": 2.539449394101112e-07,
        "max_row_sum_error": 2.5395899383440224e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "staged_stream_3.f32",
        "elements": 524288,
        "max_abs_error": 2.747770866395882e-07,
        "max_row_sum_error": 2.74786958298634e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "staged_graph_3.f32",
        "elements": 524288,
        "max_abs_error": 2.747770866395882e-07,
        "max_row_sum_error": 2.74786958298634e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "fused_stream_3.f32",
        "elements": 524288,
        "max_abs_error": 2.747770866395882e-07,
        "max_row_sum_error": 2.74786958298634e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "fused_graph_3.f32",
        "elements": 524288,
        "max_abs_error": 2.747770866395882e-07,
        "max_row_sum_error": 2.74786958298634e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "cudnn_stream_3.f32",
        "elements": 524288,
        "max_abs_error": 2.8696413112427877e-07,
        "max_row_sum_error": 2.8697785370290774e-07,
        "same_algorithm_bitwise_checked": false
      },
      {
        "file": "cudnn_graph_3.f32",
        "elements": 524288,
        "max_abs_error": 2.8696413112427877e-07,
        "max_row_sum_error": 2.8697785370290774e-07,
        "same_algorithm_bitwise_checked": false
      }
    ],
    "intermediate_checks": 0,
    "atol": 1e-07,
    "rtol": 2e-05,
    "row_sum_atol": 2e-05
  },
  "traces": [],
  "timings": {
    "samples": 147,
    "timing_variants": {
      "cudnn_graph": {
        "host_single_us": {
          "p10": 7.794,
          "median": 7.977,
          "p90": 8.258
        },
        "host_enqueue_us": {
          "p10": 1.85121875,
          "median": 1.87578125,
          "p90": 1.89296875
        },
        "host_batch_us": {
          "p10": 3.08990625,
          "median": 3.1035625,
          "p90": 3.12025
        },
        "event_interval_us": {
          "p10": 2.90700006485,
          "median": 2.92000007629,
          "p90": 2.92599987984
        }
      },
      "cudnn_stream": {
        "host_single_us": {
          "p10": 8.485,
          "median": 8.672,
          "p90": 9.018
        },
        "host_enqueue_us": {
          "p10": 1.888,
          "median": 2.0238125,
          "p90": 2.0789375
        },
        "host_batch_us": {
          "p10": 3.1984375,
          "median": 3.218375,
          "p90": 3.2313125
        },
        "event_interval_us": {
          "p10": 3.01499986649,
          "median": 3.03500008583,
          "p90": 3.04600000381
        }
      },
      "empty_sync": {
        "host_single_us": {
          "p10": 1.422,
          "median": 1.442,
          "p90": 1.49
        }
      },
      "fused_graph": {
        "host_single_us": {
          "p10": 22.446,
          "median": 22.613,
          "p90": 22.946
        },
        "host_enqueue_us": {
          "p10": 1.81709375,
          "median": 1.83184375,
          "p90": 1.86425
        },
        "host_batch_us": {
          "p10": 17.70475,
          "median": 17.71353125,
          "p90": 17.723625
        },
        "event_interval_us": {
          "p10": 17.5230007172,
          "median": 17.5310001373,
          "p90": 17.5410003662
        }
      },
      "fused_stream": {
        "host_single_us": {
          "p10": 22.994,
          "median": 23.184,
          "p90": 23.8
        },
        "host_enqueue_us": {
          "p10": 1.7024375,
          "median": 1.8378125,
          "p90": 1.911875
        },
        "host_batch_us": {
          "p10": 17.84509375,
          "median": 17.877625,
          "p90": 17.893
        },
        "event_interval_us": {
          "p10": 17.6539993286,
          "median": 17.6919994354,
          "p90": 17.7129993439
        }
      },
      "staged_graph": {
        "host_single_us": {
          "p10": 21.919,
          "median": 22.012,
          "p90": 22.186
        },
        "host_enqueue_us": {
          "p10": 1.82475,
          "median": 1.84596875,
          "p90": 1.8756875
        },
        "host_batch_us": {
          "p10": 17.075375,
          "median": 17.0838125,
          "p90": 17.09834375
        },
        "event_interval_us": {
          "p10": 16.892999649,
          "median": 16.9029998779,
          "p90": 16.9159984589
        }
      },
      "staged_stream": {
        "host_single_us": {
          "p10": 25.079,
          "median": 25.316,
          "p90": 25.799
        },
        "host_enqueue_us": {
          "p10": 6.8493125,
          "median": 7.3558125,
          "p90": 7.4853125
        },
        "host_batch_us": {
          "p10": 19.8945,
          "median": 19.92609375,
          "p90": 19.98309375
        },
        "event_interval_us": {
          "p10": 19.7089996338,
          "median": 19.7399997711,
          "p90": 19.8010005951
        }
      }
    },
    "paired_completion_ratios": {
      "cudnn_stream/cudnn_graph": {
        "p10": 1.0397652413462688,
        "median": 1.0852830188679246,
        "p90": 1.1470096463022506
      },
      "fused_stream/fused_graph": {
        "p10": 1.0168037498894489,
        "median": 1.0231694375663247,
        "p90": 1.037521621501752
      },
      "staged_stream/staged_graph": {
        "p10": 1.1339237303452014,
        "median": 1.150448114280515,
        "p90": 1.1701818841565745
      }
    }
  }
}
