{
  "schema_version": 1,
  "status": "measured",
  "cases": 256,
  "trials": 21,
  "repetitions_per_backend": 3,
  "CPU_workers": 64,
  "hardware": {
    "cpu_model": "INTEL(R) XEON(R) GOLD 6530",
    "gpu_model": "NVIDIA L40S"
  },
  "modes": {
    "fused_fp64": "FP64 complete economy",
    "mixed": "FP32 share predictor + FP64 correction; remaining economic stages FP64"
  },
  "timings": [
    {
      "backend": "fused_fp64_1GPU",
      "mode": "fused_fp64",
      "GPU_count": 1,
      "median_seconds": 23.104696768335998,
      "min_seconds": 23.10419918037951,
      "max_seconds": 23.11111296340823,
      "economies_per_second": 11.079998260389942,
      "speed_vs_same_mode_1GPU": 1.0,
      "device_scaling_efficiency": 1.0,
      "speed_vs_same_run_CPU": 1.552456669082106
    },
    {
      "backend": "fused_fp64_2GPU",
      "mode": "fused_fp64",
      "GPU_count": 2,
      "median_seconds": 12.386754082515836,
      "min_seconds": 12.334212884306908,
      "max_seconds": 12.469313933514059,
      "economies_per_second": 20.667238430231645,
      "speed_vs_same_mode_1GPU": 1.865274519411729,
      "device_scaling_efficiency": 0.9326372597058645,
      "speed_vs_same_run_CPU": 2.895757867329659
    },
    {
      "backend": "fused_fp64_4GPU",
      "mode": "fused_fp64",
      "GPU_count": 4,
      "median_seconds": 7.119610080495477,
      "min_seconds": 7.077951053157449,
      "max_seconds": 7.180769925005734,
      "economies_per_second": 35.95702532942424,
      "speed_vs_same_mode_1GPU": 3.245219402061421,
      "device_scaling_efficiency": 0.8113048505153553,
      "speed_vs_same_run_CPU": 5.038062503364897
    },
    {
      "backend": "mixed_1GPU",
      "mode": "mixed",
      "GPU_count": 1,
      "median_seconds": 17.520079816691577,
      "min_seconds": 17.499393618665636,
      "max_seconds": 17.57722275238484,
      "economies_per_second": 14.611805578425843,
      "speed_vs_same_mode_1GPU": 1.0,
      "device_scaling_efficiency": 1.0,
      "speed_vs_same_run_CPU": 2.0473103410722002
    },
    {
      "backend": "mixed_2GPU",
      "mode": "mixed",
      "GPU_count": 2,
      "median_seconds": 9.348845077678561,
      "min_seconds": 9.328364234417677,
      "max_seconds": 9.350621780380607,
      "economies_per_second": 27.383061530373343,
      "speed_vs_same_mode_1GPU": 1.874036811084053,
      "device_scaling_efficiency": 0.9370184055420265,
      "speed_vs_same_run_CPU": 3.836734942882351
    },
    {
      "backend": "mixed_4GPU",
      "mode": "mixed",
      "GPU_count": 4,
      "median_seconds": 5.604943481273949,
      "min_seconds": 5.577504698187113,
      "max_seconds": 5.635201690718532,
      "economies_per_second": 45.67396635760789,
      "speed_vs_same_mode_1GPU": 3.1258263130085004,
      "device_scaling_efficiency": 0.7814565782521251,
      "speed_vs_same_run_CPU": 6.399536535017891
    },
    {
      "backend": "CPU64",
      "mode": "CPU",
      "GPU_count": 0,
      "median_seconds": 35.869040585123,
      "min_seconds": 35.57547185290605,
      "max_seconds": 35.87261546589434,
      "economies_per_second": 7.137074084612629,
      "speed_vs_same_mode_1GPU": "",
      "device_scaling_efficiency": "",
      "speed_vs_same_run_CPU": ""
    }
  ],
  "validated_records": 5376,
  "pure_FP32_full_economy_measured": false,
  "CPU_same_node_same_allocation": true,
  "no_reestimation": true
}