{
  "schema": "aner.knn-parallel-evidence/1",
  "recorded_date": "2026-10-30",
  "build": {
    "configuration": "Release",
    "compiler": "AppleClang 21.0.0.21000334",
    "target": "macOS arm64",
    "cpu_instruction_baseline": "apple-m1",
    "lto": "thin, both original baseline and optimized",
    "language": "C++20",
    "optimization": "-O3 -DNDEBUG",
    "numerical_flags": [
      "-fno-fast-math",
      "-ffp-contract=off"
    ],
    "provider": "Apple Accelerate linked for Tensor runtime; KNN distance loop is native scalar CPU"
  },
  "scope": [
    "Same machine, sequential experiments; no builds or tests from this task ran concurrently with benchmark calls. Host CPUs were not reserved.",
    "Prediction timers exclude fitting, input construction, correctness verification and process startup. Cold context mode includes pool creation/destruction.",
    "Existing CPU Float64 inputs were materialized before prediction timing. Hardware utilization and RSS were not directly measured.",
    "Process CPU/wall ratios are timer-derived estimates; clock resolution limits their interpretation for tiny requests.",
    "Each dataset retains all samples and median/MAD, including exploratory runs that exposed cancellation-token reference-count contention.",
    "Native query parallelism preserves total exhaustive distance work. Weighting is a deliberately changed estimator, with no original weighted baseline.",
    "Local CPU results do not establish GPU, distributed or other-platform speedups."
  ],
  "validation": {
    "native_checks": 107427,
    "language_checks": 165,
    "selected_ctest_suites": 16,
    "asan_ubsan": "pass; macOS leak detector unavailable, disabled",
    "decimal_reference": {
      "precision_digits": 1100,
      "cases": 39,
      "probabilities": 132,
      "maximum_observed_ulps": {
        "normal": 2,
        "subnormal": 0
      },
      "test_bounds_ulps": {
        "normal": 4,
        "subnormal": 1
      }
    },
    "worker_determinism": "exact successful labels, probabilities, indices and distances across admitted 1/2/4 worker contexts in the same numerical environment",
    "integration_scope": "15/16 selected suites passed in sandbox; Decimal transport then passed 2196 checks with approved host shared-memory access. Full integrated editor suite and ASan/UBSan passed.",
    "kernel_sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64",
    "primary_measurements": "integrated-* groups; earlier complete groups retained as exploratory history"
  },
  "experiments": [
    {
      "name": "batch",
      "kernel": "exploratory-before-token-borrow",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:15:56Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "1d92e6d61f5f0817365d9a7c7c1b45a843cf7c594ea0f79b9642da54ed6d21c4"
          },
          "optimized": {
            "sha256": "fba10775d01c62f49147ecfe069e3428c5a41acbe673bae6b866aa8c0ffc3c69"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "ac7a04f65df59b8634de88ef46b3bdd49e850b48c723d4c5b9ce4f0cc6b9e300"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "932a723d3e3197a5f174fe6eac77fb4f3d029b7358ef26f5c782610d2585df93"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.058286125,
            "mad_seconds": 0.0005082920000000005,
            "rows_per_second": 4392.125913328429,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.058245625,
              0.058225291,
              0.058794417,
              0.057267125,
              0.058286125,
              0.089163583,
              0.066732041
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9864922080014367
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.058351167,
            "mad_seconds": 0.00021416699999999428,
            "rows_per_second": 4387.230164565517,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.05793375,
              0.058565334,
              0.059317208,
              0.058351167,
              0.058267167,
              0.057558208,
              0.058497083
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.970178625000699,
            "speedup_vs_original_serial": 0.9988853350610793
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.065539709,
            "mad_seconds": 0.0025477089999999952,
            "rows_per_second": 3906.0289388834485,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.0648455,
              0.070536333,
              0.068671583,
              0.065395792,
              0.065539709,
              0.062992,
              0.069452125
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0935841249956866,
            "speedup_vs_optimized_serial": 0.8903177614047691,
            "speedup_vs_original_serial": 0.8893253554116329
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.088699041,
            "mad_seconds": 0.0033218339999999957,
            "rows_per_second": 2886.164237108268,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.08810025,
              0.090389541,
              0.088699041,
              0.077176958,
              0.092020875,
              0.085235083,
              0.1117945
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.269742166994547,
            "speedup_vs_optimized_serial": 0.6578556694880162,
            "speedup_vs_original_serial": 0.6571223808383677
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.063277709,
            "mad_seconds": 0.003486751000000003,
            "rows_per_second": 4045.658479828971,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.059790958,
              0.075281375,
              0.065324916,
              0.068480334,
              0.054134666,
              0.062681458,
              0.063277709
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.3516701249973266,
            "speedup_vs_optimized_serial": 0.9221441155526032,
            "speedup_vs_original_serial": 0.9211162338383648
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.052669792,
            "mad_seconds": 0.0015604170000000014,
            "rows_per_second": 4860.471064704413,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056086125,
              0.052669792,
              0.049845125,
              0.051109375,
              0.049812291,
              0.052903416,
              0.053495958
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0286514999970677,
            "speedup_vs_optimized_serial": 1.1078678077938868,
            "speedup_vs_original_serial": 1.1066329063915803
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.043343833,
            "mad_seconds": 0.0019960829999999957,
            "rows_per_second": 5906.261220598557,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.043343833,
              0.037780459,
              0.04134775,
              0.044156708,
              0.043883375,
              0.0387385,
              0.048784667
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8401170409997576,
            "speedup_vs_optimized_serial": 1.3462391985498838,
            "speedup_vs_original_serial": 1.3447385929158597
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.048634958,
            "mad_seconds": 0.0008637079999999978,
            "rows_per_second": 5263.703527820462,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.049466875,
              0.048634958,
              0.047046166,
              0.047172292,
              0.04777125,
              0.049450792,
              0.050933542
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7032613339979434,
            "speedup_vs_optimized_serial": 1.1997782952747693,
            "speedup_vs_original_serial": 1.1984409444745485
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057602542,
            "mad_seconds": 5.079200000000089e-05,
            "rows_per_second": 4444.24831112488,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057637916,
              0.058216042,
              0.057766,
              0.057483125,
              0.05755175,
              0.057602542,
              0.057575916
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7799739170004614
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.058980167,
            "mad_seconds": 0.00029024999999999884,
            "rows_per_second": 4340.4421014948975,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.072241541,
              0.088314334,
              0.058980167,
              0.0580695,
              0.058992917,
              0.058689917,
              0.058723375
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.1105283750002855,
            "speedup_vs_optimized_serial": 0.9766425720700316
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.059659,
            "mad_seconds": 0.00032962500000000006,
            "rows_per_second": 4291.054157796812,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.059329375,
              0.059029083,
              0.058761042,
              0.059659,
              0.059676,
              0.059942208,
              0.060115625
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7870539159994223,
            "speedup_vs_optimized_serial": 0.9655297943311152
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.051061667,
            "mad_seconds": 0.001300084,
            "rows_per_second": 5013.545679971631,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.051061667,
              0.051586833,
              0.051270084,
              0.049451041,
              0.053712292,
              0.049761583,
              0.049119375
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9363458750012796,
            "speedup_vs_optimized_serial": 1.1280975609354862
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.06024425,
            "mad_seconds": 0.0026651659999999966,
            "rows_per_second": 4249.368196964856,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.053691917,
              0.0647035,
              0.06024425,
              0.064186541,
              0.061358334,
              0.059078625,
              0.057579084
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.800926541996887,
            "speedup_vs_optimized_serial": 0.9561500392153608
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.059095917,
            "mad_seconds": 0.005776375,
            "rows_per_second": 4331.940563677183,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.061250833,
              0.050527792,
              0.064872292,
              0.059095917,
              0.060882291,
              0.051385917,
              0.051241541
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.990238500002306,
            "speedup_vs_optimized_serial": 0.974729641643432
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.045075708,
            "mad_seconds": 0.0019552089999999994,
            "rows_per_second": 5679.333977405302,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.045016041,
              0.047030917,
              0.045075708,
              0.046381292,
              0.042601125,
              0.041445084,
              0.048027333
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.6788337500038324,
            "speedup_vs_optimized_serial": 1.2779065389277968
          },
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057298459,
            "mad_seconds": 0.00022924899999999387,
            "rows_per_second": 4467.833942968693,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.05701025,
              0.057093167,
              0.057751542,
              0.057298459,
              0.05713975,
              0.059071375,
              0.057527708
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.768901959003415
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.059487875,
            "mad_seconds": 0.0005969580000000016,
            "rows_per_second": 4303.397961349267,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.060130542,
              0.059359292,
              0.058890917,
              0.062342166,
              0.059639583,
              0.058634125,
              0.059487875
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7965149579977151,
            "speedup_vs_original_serial": 0.9631955923791193
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.062024916,
            "mad_seconds": 0.00021120900000000387,
            "rows_per_second": 4127.373586447098,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.061775917,
              0.062024916,
              0.065360584,
              0.062195208,
              0.062236125,
              0.061928708,
              0.06140525
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.1387157499993918,
            "speedup_vs_optimized_serial": 0.9590964218315104,
            "speedup_vs_original_serial": 0.9237974461746954
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.06291925,
            "mad_seconds": 0.000956499999999999,
            "rows_per_second": 4068.7071126880887,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.06657475,
              0.06387575,
              0.063367416,
              0.061062084,
              0.06291925,
              0.061085208,
              0.062465417
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8290323750043171,
            "speedup_vs_optimized_serial": 0.9454638286374998,
            "speedup_vs_original_serial": 0.9106665924975267
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.059314542,
            "mad_seconds": 0.0024586259999999985,
            "rows_per_second": 4315.973644372067,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.061400875,
              0.060547125,
              0.065899208,
              0.056855916,
              0.059314542,
              0.055715166,
              0.050841875
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0163844589988003,
            "speedup_vs_optimized_serial": 1.002922268201953,
            "speedup_vs_original_serial": 0.9660103082309901
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057375,
            "mad_seconds": 0.0006362500000000049,
            "rows_per_second": 4461.8736383442265,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057375,
              0.056933792,
              0.056063333,
              0.057581667,
              0.05673875,
              0.0593595,
              0.066950208
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7898060840016115,
            "speedup_vs_optimized_serial": 1.0368257080610022,
            "speedup_vs_original_serial": 0.9986659520697168
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.053440334,
            "mad_seconds": 0.0008785830000000022,
            "rows_per_second": 4790.389221744011,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056339334,
              0.0533465,
              0.044070208,
              0.058587209,
              0.054318917,
              0.052965916,
              0.053440334
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9967185419955058,
            "speedup_vs_optimized_serial": 1.1131643563455274,
            "speedup_vs_original_serial": 1.0721950016255513
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.061199,
            "mad_seconds": 0.0035503750000000014,
            "rows_per_second": 4183.074886844556,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057648625,
              0.0671815,
              0.064051,
              0.061806583,
              0.061199,
              0.053505666,
              0.056610459
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8050842919983552,
            "speedup_vs_optimized_serial": 0.9720399843134692,
            "speedup_vs_original_serial": 0.9362646285070018
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.058918792,
            "mad_seconds": 0.001123291999999998,
            "rows_per_second": 4344.963488049789,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.059557542,
              0.058040625,
              0.057774084,
              0.0577955,
              0.062683625,
              0.061325542,
              0.058918792
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7996062910024193
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.062152375,
            "mad_seconds": 0.0012769590000000011,
            "rows_per_second": 4118.909373937842,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.060558458,
              0.06083925,
              0.062152375,
              0.063211375,
              0.060875416,
              0.063389791,
              0.063720166
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0944742920037243,
            "speedup_vs_optimized_serial": 0.9479732994917731
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.061205,
            "mad_seconds": 0.0002124579999999987,
            "rows_per_second": 4182.664814966098,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.061107834,
              0.060756875,
              0.060724209,
              0.061205,
              0.061315708,
              0.061417458,
              0.061631875
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.810065666999435,
            "speedup_vs_optimized_serial": 0.9626467118699452
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.063609833,
            "mad_seconds": 0.0014694589999999924,
            "rows_per_second": 4024.5350117488906,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.065384875,
              0.063511875,
              0.064678042,
              0.065079292,
              0.059256583,
              0.063609833,
              0.060997458
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0743546250014333,
            "speedup_vs_optimized_serial": 0.926252895523244
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.0564985,
            "mad_seconds": 0.0024119579999999988,
            "rows_per_second": 4531.09374585166,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.053116583,
              0.0564985,
              0.059753334,
              0.058383083,
              0.058910458,
              0.056268792,
              0.052386833
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7699869170028251,
            "speedup_vs_optimized_serial": 1.042838163845058
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.054536167,
            "mad_seconds": 0.0044111670000000006,
            "rows_per_second": 4694.132611116582,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.063137084,
              0.050125,
              0.052149333,
              0.056586417,
              0.054536167,
              0.038881875,
              0.062507042
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9881634170014877,
            "speedup_vs_optimized_serial": 1.080361808339042
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.049997958,
            "mad_seconds": 0.0025151659999999992,
            "rows_per_second": 5120.209109340025,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.047482792,
              0.048044583,
              0.052578167,
              0.048057875,
              0.056547958,
              0.049997958,
              0.05297425
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7213627919991268,
            "speedup_vs_optimized_serial": 1.1784239668348055
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "batch-cpu",
      "kernel": "exploratory-before-token-borrow",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:18:37Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "089073b5229a3f8a480c178d71c21e06f2d0a2704073e495cabc8abf5223943d"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "ac7a04f65df59b8634de88ef46b3bdd49e850b48c723d4c5b9ce4f0cc6b9e300"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.056974125,
            "mad_seconds": 0.00014170799999999734,
            "median_process_cpu_seconds": 0.056901,
            "median_process_cpu_wall_ratio": 0.9986722945982421,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4493.267777258537,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057171416,
              0.057115833,
              0.056974125,
              0.056872625,
              0.0569335,
              0.056835167,
              0.057387083,
              0.05653125,
              0.057375834
            ],
            "raw_process_cpu_seconds": [
              0.057104,
              0.05704,
              0.056901,
              0.056812,
              0.056843,
              0.056753,
              0.057327,
              0.056412,
              0.05727
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.2537289160027285
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057258166,
            "mad_seconds": 0.00026995900000000017,
            "median_process_cpu_seconds": 0.057172,
            "median_process_cpu_wall_ratio": 0.9985989195109514,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4470.977991156755,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.058679292,
              0.057217041,
              0.056966166,
              0.057528125,
              0.05747875,
              0.057396417,
              0.057258166,
              0.0564375,
              0.056955666
            ],
            "raw_process_cpu_seconds": [
              0.058617,
              0.057137,
              0.056883,
              0.057441,
              0.057413,
              0.057316,
              0.057172,
              0.056354,
              0.056876
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0672170420002658,
            "speedup_vs_original_serial": 0.9950392927359916
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.059508833,
            "mad_seconds": 7.5376000000002e-05,
            "median_process_cpu_seconds": 0.059433,
            "median_process_cpu_wall_ratio": 0.9990493235017514,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4301.88237769677,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.059233208,
              0.059296792,
              0.059508833,
              0.059439125,
              0.060629208,
              0.059536458,
              0.059687125,
              0.059584209,
              0.059475542
            ],
            "raw_process_cpu_seconds": [
              0.059191,
              0.059262,
              0.059433,
              0.059392,
              0.060589,
              0.059457,
              0.059599,
              0.059509,
              0.059419
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.1463296250003623,
            "speedup_vs_optimized_serial": 0.9621792784946732,
            "speedup_vs_original_serial": 0.9574061887585663
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.059708583,
            "mad_seconds": 0.00010145800000000565,
            "median_process_cpu_seconds": 0.059605,
            "median_process_cpu_wall_ratio": 0.9988276824593659,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4287.490795083849,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.059814625,
              0.059607125,
              0.070066208,
              0.059674958,
              0.059815709,
              0.059717417,
              0.059299167,
              0.059640834,
              0.059708583
            ],
            "raw_process_cpu_seconds": [
              0.059777,
              0.059552,
              0.068519,
              0.059605,
              0.059737,
              0.059643,
              0.059242,
              0.059571,
              0.059578
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9221055410016561,
            "speedup_vs_optimized_serial": 0.9589603893296211,
            "speedup_vs_original_serial": 0.9542032675603773
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.05394225,
            "mad_seconds": 0.0021392909999999946,
            "median_process_cpu_seconds": 0.103425,
            "median_process_cpu_wall_ratio": 1.9478191360294856,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4745.8161274325785,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.05195,
              0.051802959,
              0.055963542,
              0.057582667,
              0.061162042,
              0.06200725,
              0.053052667,
              0.050835125,
              0.05394225
            ],
            "raw_process_cpu_seconds": [
              0.103271,
              0.103425,
              0.102368,
              0.110844,
              0.105076,
              0.10982,
              0.103337,
              0.101096,
              0.107702
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0789684999981546,
            "speedup_vs_optimized_serial": 1.0614715923047333,
            "speedup_vs_original_serial": 1.0562059424662487
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.054151833,
            "mad_seconds": 0.0015168759999999934,
            "median_process_cpu_seconds": 0.107794,
            "median_process_cpu_wall_ratio": 1.9933693561681471,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4727.44846882653,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.05215375,
              0.054151833,
              0.054595,
              0.05112975,
              0.055668709,
              0.052256208,
              0.054402209,
              0.0568545,
              0.053403709
            ],
            "raw_process_cpu_seconds": [
              0.103953,
              0.107794,
              0.108828,
              0.101954,
              0.111057,
              0.104008,
              0.108533,
              0.113026,
              0.106531
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8494346669976949,
            "speedup_vs_optimized_serial": 1.057363395252013,
            "speedup_vs_original_serial": 1.0521181249764897
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.049026041,
            "mad_seconds": 0.0006811249999999977,
            "median_process_cpu_seconds": 0.192118,
            "median_process_cpu_wall_ratio": 3.933586814894963,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 5221.714720958194,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.048711875,
              0.049786,
              0.049214625,
              0.050060833,
              0.03827275,
              0.048344916,
              0.049026041,
              0.048765417,
              0.051682917
            ],
            "raw_process_cpu_seconds": [
              0.191818,
              0.196203,
              0.19359,
              0.199149,
              0.150771,
              0.189981,
              0.192118,
              0.190971,
              0.202958
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0071129169955384,
            "speedup_vs_optimized_serial": 1.167913313661203,
            "speedup_vs_original_serial": 1.1621196376023917
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.04767825,
            "mad_seconds": 0.0015446250000000009,
            "median_process_cpu_seconds": 0.187787,
            "median_process_cpu_wall_ratio": 3.93851869172457,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 5369.324587206956,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.045871083,
              0.047527958,
              0.046133625,
              0.045568084,
              0.04767825,
              0.0497365,
              0.048385958,
              0.048639542,
              0.051015083
            ],
            "raw_process_cpu_seconds": [
              0.180137,
              0.187787,
              0.179555,
              0.179902,
              0.185585,
              0.196361,
              0.190569,
              0.18988,
              0.202105
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7868897919979645,
            "speedup_vs_optimized_serial": 1.200928431727255,
            "speedup_vs_original_serial": 1.1949709773324315
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.05827,
            "mad_seconds": 0.00013904199999999894,
            "median_process_cpu_seconds": 0.058138,
            "median_process_cpu_wall_ratio": 0.9982976935248035,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4393.341342028488,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.0579245,
              0.05829725,
              0.058202209,
              0.05830325,
              0.058419083,
              0.05827,
              0.058130958,
              0.05858275,
              0.057875875
            ],
            "raw_process_cpu_seconds": [
              0.057822,
              0.057843,
              0.058138,
              0.058204,
              0.058342,
              0.058208,
              0.058026,
              0.058498,
              0.057776
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8960269160015741
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.059845708,
            "mad_seconds": 0.00010287500000000227,
            "median_process_cpu_seconds": 0.059784,
            "median_process_cpu_wall_ratio": 0.9987392028932528,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4277.66682950764,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.059641292,
              0.059941625,
              0.05984025,
              0.059948583,
              0.05975725,
              0.059726834,
              0.059845708,
              0.060393666,
              0.060461917
            ],
            "raw_process_cpu_seconds": [
              0.059575,
              0.059854,
              0.059784,
              0.059873,
              0.059683,
              0.059662,
              0.059757,
              0.060242,
              0.060365
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.153236582998943,
            "speedup_vs_optimized_serial": 0.9736704927945711
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.06022,
            "mad_seconds": 0.0002043749999999997,
            "median_process_cpu_seconds": 0.060149,
            "median_process_cpu_wall_ratio": 0.9984875578429951,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4251.079375622717,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.060849458,
              0.059714625,
              0.060604834,
              0.0611025,
              0.060246042,
              0.06022,
              0.060057834,
              0.060015625,
              0.060203041
            ],
            "raw_process_cpu_seconds": [
              0.060791,
              0.059634,
              0.060432,
              0.060938,
              0.060153,
              0.060149,
              0.059967,
              0.059935,
              0.060111
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9162621659997967,
            "speedup_vs_optimized_serial": 0.9676187313184988
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.051088542,
            "mad_seconds": 0.0010764580000000024,
            "median_process_cpu_seconds": 0.101538,
            "median_process_cpu_wall_ratio": 1.991938588281984,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 5010.908316780698,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.050090167,
              0.051364084,
              0.0551675,
              0.050576333,
              0.051088542,
              0.053299417,
              0.050012084,
              0.054654667,
              0.049840625
            ],
            "raw_process_cpu_seconds": [
              0.099846,
              0.102317,
              0.109374,
              0.100922,
              0.101538,
              0.106212,
              0.099621,
              0.10443,
              0.09857
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0457307499964372,
            "speedup_vs_optimized_serial": 1.1405688578859816
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.052143875,
            "mad_seconds": 0.001967166999999999,
            "median_process_cpu_seconds": 0.104088,
            "median_process_cpu_wall_ratio": 1.9922150916079466,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4909.493205098393,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.054111042,
              0.054664125,
              0.049085583,
              0.051285125,
              0.051297875,
              0.054153917,
              0.051925875,
              0.052143875,
              0.055577708
            ],
            "raw_process_cpu_seconds": [
              0.107534,
              0.109069,
              0.097698,
              0.102171,
              0.101925,
              0.108037,
              0.103342,
              0.104088,
              0.110962
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8695237499996438,
            "speedup_vs_optimized_serial": 1.117485035394857
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.047513625,
            "mad_seconds": 0.0021244170000000034,
            "median_process_cpu_seconds": 0.185811,
            "median_process_cpu_wall_ratio": 3.9339618035540975,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 5387.928199542763,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.047513625,
              0.049147375,
              0.049638042,
              0.043237583,
              0.042437625,
              0.048733583,
              0.044910667,
              0.052084125,
              0.046824959
            ],
            "raw_process_cpu_seconds": [
              0.185811,
              0.194099,
              0.197232,
              0.170095,
              0.167106,
              0.192052,
              0.175324,
              0.202328,
              0.182957
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9754079999984242,
            "speedup_vs_optimized_serial": 1.2263850632318625
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.048538875,
            "mad_seconds": 0.0012109999999999968,
            "median_process_cpu_seconds": 0.189672,
            "median_process_cpu_wall_ratio": 3.9103869506608713,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 5274.12306115459,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.048308667,
              0.049749875,
              0.048538875,
              0.049823458,
              0.050469917,
              0.049509959,
              0.047494792,
              0.046210833,
              0.040295167
            ],
            "raw_process_cpu_seconds": [
              0.188184,
              0.193736,
              0.189672,
              0.194829,
              0.197421,
              0.194438,
              0.186238,
              0.180138,
              0.157689
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7891498749959283,
            "speedup_vs_optimized_serial": 1.2004810577088982
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "tiny",
      "kernel": "exploratory-before-token-borrow",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:18:52Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "089073b5229a3f8a480c178d71c21e06f2d0a2704073e495cabc8abf5223943d"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "ac7a04f65df59b8634de88ef46b3bdd49e850b48c723d4c5b9ce4f0cc6b9e300"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.09e-07,
            "mad_seconds": 9.999999999999692e-10,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.4124293785310733,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1410437.2355430182,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              7.08e-07,
              1.584e-06,
              7.92e-07,
              7.5e-07,
              7.5e-07,
              7.08e-07,
              7.09e-07,
              6.66e-07,
              7.5e-07,
              7.91e-07,
              7.5e-07,
              7.09e-07,
              7.08e-07,
              7.09e-07,
              7.08e-07,
              7.09e-07,
              7.08e-07,
              7.5e-07,
              7.08e-07,
              7.5e-07,
              8.34e-07,
              7.08e-07,
              6.66e-07,
              7.09e-07,
              7.08e-07,
              7.08e-07,
              7.5e-07,
              6.67e-07,
              7.08e-07,
              7.09e-07,
              7.08e-07
            ],
            "raw_process_cpu_seconds": [
              3e-06,
              2e-06,
              1e-06,
              2e-06,
              1e-06,
              0,
              1e-06,
              0,
              2e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005148874995938968
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 8.33e-07,
            "mad_seconds": 4.100000000000001e-08,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.2004801920768307,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1200480.1920768307,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              8.75e-07,
              8.33e-07,
              8.34e-07,
              8.75e-07,
              8.34e-07,
              8.33e-07,
              8.34e-07,
              8.33e-07,
              7.91e-07,
              8.33e-07,
              7.92e-07,
              7.92e-07,
              7.92e-07,
              8.33e-07,
              7.91e-07,
              8.33e-07,
              7.92e-07,
              8.75e-07,
              8.33e-07,
              8.75e-07,
              9.17e-07,
              8.34e-07,
              8.33e-07,
              7.91e-07,
              7.91e-07,
              7.91e-07,
              8.75e-07,
              7.92e-07,
              7.91e-07,
              7.91e-07,
              7.91e-07
            ],
            "raw_process_cpu_seconds": [
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              0,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              0,
              0,
              0,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.004937625002639834,
            "speedup_vs_original_serial": 0.851140456182473
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 8.292e-06,
            "mad_seconds": 2.5419999999999995e-06,
            "median_process_cpu_seconds": 1.2e-05,
            "median_process_cpu_wall_ratio": 1.377004032654667,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 120598.16690786301,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.875e-06,
              6.416e-06,
              4.333e-06,
              1.1792e-05,
              9.625e-06,
              1.0291e-05,
              1.2875e-05,
              8.583e-06,
              1.1416e-05,
              1.2292e-05,
              1.0167e-05,
              1.35e-05,
              8.375e-06,
              5.75e-06,
              4.583e-06,
              4.541e-06,
              4.25e-06,
              9.5e-06,
              8.167e-06,
              9.75e-06,
              8.292e-06,
              5.334e-06,
              1.2834e-05,
              7.583e-06,
              6e-06,
              1.0708e-05,
              4.125e-06,
              5.416e-06,
              8.708e-06,
              5.417e-06,
              8.084e-06
            ],
            "raw_process_cpu_seconds": [
              1e-05,
              9e-06,
              8e-06,
              1.6e-05,
              1.2e-05,
              1.2e-05,
              1.5e-05,
              1.3e-05,
              1.5e-05,
              1.6e-05,
              1.4e-05,
              2e-05,
              1.3e-05,
              7e-06,
              5e-06,
              5e-06,
              8e-06,
              1.4e-05,
              1.3e-05,
              1.1e-05,
              1.3e-05,
              9e-06,
              1.6e-05,
              8e-06,
              9e-06,
              1.3e-05,
              7e-06,
              7e-06,
              1.4e-05,
              8e-06,
              1.1e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005230041999311652,
            "speedup_vs_optimized_serial": 0.10045827303424988,
            "speedup_vs_original_serial": 0.08550410033767487
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 3.0625e-05,
            "mad_seconds": 4.4169999999999985e-06,
            "median_process_cpu_seconds": 3.7e-05,
            "median_process_cpu_wall_ratio": 1.2018779342723005,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 32653.061224489797,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              3.6625e-05,
              2.9167e-05,
              2.4084e-05,
              3.875e-05,
              3.4541e-05,
              3.075e-05,
              3.1291e-05,
              2.6958e-05,
              3.5375e-05,
              5.5958e-05,
              3.125e-05,
              2.7e-05,
              4.9875e-05,
              3.0625e-05,
              3.3958e-05,
              2.6208e-05,
              2.3e-05,
              2.6166e-05,
              2.8709e-05,
              4.2916e-05,
              3.0709e-05,
              2.5e-05,
              2.45e-05,
              3.6125e-05,
              3.2083e-05,
              2.9667e-05,
              2.55e-05,
              2.3583e-05,
              2.6625e-05,
              2.5708e-05,
              3.3041e-05
            ],
            "raw_process_cpu_seconds": [
              4.3e-05,
              3.7e-05,
              3.1e-05,
              4.6e-05,
              5.1e-05,
              3.9e-05,
              3.9e-05,
              3.3e-05,
              4.2e-05,
              6.4e-05,
              3.8e-05,
              3.1e-05,
              6e-05,
              3.4e-05,
              3.7e-05,
              3.1e-05,
              2.8e-05,
              3.3e-05,
              3.2e-05,
              5e-05,
              3.6e-05,
              2.9e-05,
              2.9e-05,
              4.1e-05,
              4e-05,
              3.7e-05,
              3.2e-05,
              2.9e-05,
              3.2e-05,
              2.9e-05,
              4e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005877874995348975,
            "speedup_vs_optimized_serial": 0.027200000000000002,
            "speedup_vs_original_serial": 0.023151020408163265
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.459e-06,
            "mad_seconds": 1.5009999999999997e-06,
            "median_process_cpu_seconds": 1.4e-05,
            "median_process_cpu_wall_ratio": 1.8823529411764706,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 134066.2287169862,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1.1334e-05,
              3.167e-06,
              5.5e-06,
              1.4917e-05,
              1.7541e-05,
              3.2417e-05,
              7.375e-06,
              7.042e-06,
              7.083e-06,
              6.375e-06,
              1.0625e-05,
              7.084e-06,
              7.459e-06,
              2.875e-06,
              8.875e-06,
              5.417e-06,
              8.75e-06,
              6.875e-06,
              8.208e-06,
              5.958e-06,
              5.709e-06,
              1.1209e-05,
              5e-06,
              2.475e-05,
              6.125e-06,
              8.25e-06,
              1.8333e-05,
              8.792e-06,
              6.583e-06,
              8.667e-06,
              9.75e-06
            ],
            "raw_process_cpu_seconds": [
              2.7e-05,
              9e-06,
              1e-05,
              2.9e-05,
              2.8e-05,
              5.2e-05,
              1.4e-05,
              1.4e-05,
              1.4e-05,
              1.2e-05,
              2e-05,
              1.3e-05,
              1.4e-05,
              7e-06,
              1.6e-05,
              1.1e-05,
              1.6e-05,
              1.3e-05,
              1.7e-05,
              1.1e-05,
              1.2e-05,
              2e-05,
              9e-06,
              3.1e-05,
              1.1e-05,
              1.6e-05,
              3.2e-05,
              1.8e-05,
              1.3e-05,
              1.7e-05,
              1.8e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.00539720799861243,
            "speedup_vs_optimized_serial": 0.1116771685212495,
            "speedup_vs_original_serial": 0.0950529561603432
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 4.0208e-05,
            "mad_seconds": 2.958000000000001e-06,
            "median_process_cpu_seconds": 6.1e-05,
            "median_process_cpu_wall_ratio": 1.516602809706258,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 24870.672502984482,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.2625e-05,
              3.725e-05,
              4.9791e-05,
              3.6291e-05,
              3.6375e-05,
              4.375e-05,
              6.7334e-05,
              5.425e-05,
              7.2417e-05,
              6.4542e-05,
              3.8083e-05,
              4.225e-05,
              4.1459e-05,
              4.0875e-05,
              3.9458e-05,
              4.7084e-05,
              4.2167e-05,
              5.2583e-05,
              3.8625e-05,
              4.0625e-05,
              3.8375e-05,
              3.8541e-05,
              3.6333e-05,
              3.7584e-05,
              3.9375e-05,
              3.6417e-05,
              4.0208e-05,
              4.7333e-05,
              3.65e-05,
              3.5667e-05,
              3.8959e-05
            ],
            "raw_process_cpu_seconds": [
              7.2e-05,
              6.3e-05,
              7.2e-05,
              5.6e-05,
              5.6e-05,
              6.4e-05,
              9.1e-05,
              7.2e-05,
              0.000114,
              0.000112,
              5.3e-05,
              6.6e-05,
              5.8e-05,
              6e-05,
              6.6e-05,
              6.6e-05,
              6.3e-05,
              8.2e-05,
              5.9e-05,
              6.1e-05,
              5.8e-05,
              5.4e-05,
              5.3e-05,
              5.7e-05,
              5.7e-05,
              5.9e-05,
              6.1e-05,
              7.4e-05,
              5.8e-05,
              5.5e-05,
              5.5e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006696750002447516,
            "speedup_vs_optimized_serial": 0.020717270194986072,
            "speedup_vs_original_serial": 0.017633306804615997
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 1.9125e-05,
            "mad_seconds": 7.25e-06,
            "median_process_cpu_seconds": 5.7e-05,
            "median_process_cpu_wall_ratio": 2.888524356322659,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 52287.5816993464,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1.2833e-05,
              1.5666e-05,
              3.3042e-05,
              3.5875e-05,
              1.3625e-05,
              2.8042e-05,
              2.5917e-05,
              8.666e-06,
              1.7208e-05,
              3.9459e-05,
              1.75e-05,
              1.9209e-05,
              3.2125e-05,
              7.5e-06,
              4.042e-06,
              1.9208e-05,
              4.6667e-05,
              2.8667e-05,
              1.525e-05,
              1.6292e-05,
              1.9125e-05,
              2.65e-05,
              2.6375e-05,
              1.125e-05,
              1.1e-05,
              1.4875e-05,
              2.1042e-05,
              6.584e-06,
              2.1458e-05,
              7.833e-06,
              2.3417e-05
            ],
            "raw_process_cpu_seconds": [
              4.1e-05,
              4.8e-05,
              9.5e-05,
              8.5e-05,
              2.8e-05,
              8.1e-05,
              6.3e-05,
              2.6e-05,
              4.9e-05,
              0.0001,
              5.7e-05,
              5.9e-05,
              8.2e-05,
              1.9e-05,
              2.7e-05,
              6.2e-05,
              0.000123,
              7.7e-05,
              5e-05,
              5.8e-05,
              5.6e-05,
              7.1e-05,
              7.6e-05,
              3.1e-05,
              3.7e-05,
              4.8e-05,
              6.7e-05,
              2.8e-05,
              5.3e-05,
              3.5e-05,
              6.5e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005738917003327515,
            "speedup_vs_optimized_serial": 0.043555555555555556,
            "speedup_vs_original_serial": 0.0370718954248366
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.7708e-05,
            "mad_seconds": 1.074999999999999e-05,
            "median_process_cpu_seconds": 0.000149,
            "median_process_cpu_wall_ratio": 1.9759036144578312,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 12868.687908580841,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              8.1208e-05,
              9.2125e-05,
              8.0708e-05,
              7.6083e-05,
              0.00010375,
              9.5625e-05,
              9.5875e-05,
              8.1125e-05,
              9.9334e-05,
              6.6916e-05,
              6.3667e-05,
              6.3708e-05,
              7.4916e-05,
              8.475e-05,
              7.7708e-05,
              7.5542e-05,
              9.925e-05,
              8.0959e-05,
              6.8584e-05,
              8.0834e-05,
              0.000100958,
              6.6958e-05,
              8.7083e-05,
              6.775e-05,
              7e-05,
              6.4125e-05,
              6.4417e-05,
              6.3625e-05,
              7.2666e-05,
              6.6375e-05,
              0.000111958
            ],
            "raw_process_cpu_seconds": [
              0.000149,
              0.000167,
              0.000167,
              0.000148,
              0.000205,
              0.000171,
              0.000207,
              0.000157,
              0.00026,
              0.000109,
              0.000137,
              0.000128,
              0.000158,
              0.000171,
              0.000144,
              0.000146,
              0.000187,
              0.000158,
              0.000128,
              0.000168,
              0.000224,
              0.000134,
              0.000176,
              0.00014,
              0.000129,
              0.000135,
              0.000135,
              0.000116,
              0.000122,
              0.000114,
              0.000225
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.007615750000695698,
            "speedup_vs_optimized_serial": 0.010719617027847841,
            "speedup_vs_original_serial": 0.009123899727183816
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 9.59e-07,
            "mad_seconds": 1.0000000000000751e-09,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.04384133611691,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1042752.8675703858,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              9.59e-07,
              1e-06,
              1.042e-06,
              9.58e-07,
              1e-06,
              1.833e-06,
              1.042e-06,
              1e-06,
              9.58e-07,
              1.042e-06,
              9.58e-07,
              1.041e-06,
              9.58e-07,
              9.58e-07,
              9.58e-07,
              9.59e-07,
              9.58e-07,
              1.042e-06,
              9.58e-07,
              9.58e-07,
              9.59e-07,
              9.58e-07,
              1e-06,
              9.59e-07,
              1e-06,
              9.59e-07,
              9.58e-07,
              9.58e-07,
              9.58e-07,
              9.58e-07,
              9.59e-07
            ],
            "raw_process_cpu_seconds": [
              2e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06,
              3e-06,
              1e-06,
              1e-06,
              3e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.004808416997548193
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.625e-06,
            "mad_seconds": 2.8749999999999996e-06,
            "median_process_cpu_seconds": 1e-05,
            "median_process_cpu_wall_ratio": 1.2903225806451613,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 131147.54098360657,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              8.917e-06,
              1.2875e-05,
              7.583e-06,
              6e-06,
              4.959e-06,
              9.584e-06,
              4.75e-06,
              6.459e-06,
              3.958e-06,
              4.625e-06,
              7.459e-06,
              1.9583e-05,
              1.6459e-05,
              2.0417e-05,
              1.3e-05,
              7.583e-06,
              1.3958e-05,
              1.0542e-05,
              5.584e-06,
              1.1334e-05,
              7.459e-06,
              1.7792e-05,
              1.9708e-05,
              8.25e-06,
              4.5e-06,
              7.75e-06,
              7.625e-06,
              1.3458e-05,
              6.75e-06,
              7.083e-06,
              4.25e-06
            ],
            "raw_process_cpu_seconds": [
              1.3e-05,
              1.6e-05,
              1e-05,
              9e-06,
              8e-06,
              1.2e-05,
              7e-06,
              8e-06,
              6e-06,
              7e-06,
              9e-06,
              1.9e-05,
              1.8e-05,
              2.7e-05,
              1.7e-05,
              1.1e-05,
              2.1e-05,
              1.4e-05,
              6e-06,
              1.3e-05,
              1e-05,
              1.8e-05,
              2.5e-05,
              1e-05,
              8e-06,
              1e-05,
              9e-06,
              1.6e-05,
              8e-06,
              8e-06,
              6e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005289875000016764,
            "speedup_vs_optimized_serial": 0.1257704918032787
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 3.9709e-05,
            "mad_seconds": 4.832999999999999e-06,
            "median_process_cpu_seconds": 4.3e-05,
            "median_process_cpu_wall_ratio": 1.089783281733746,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 25183.207837014277,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.5792e-05,
              3.8458e-05,
              3.7125e-05,
              4.1917e-05,
              5.375e-05,
              4.1125e-05,
              3.9709e-05,
              5.6167e-05,
              5.4542e-05,
              4.0375e-05,
              4.4083e-05,
              3.6125e-05,
              4.0416e-05,
              4.4542e-05,
              4.9375e-05,
              3.9791e-05,
              3.7542e-05,
              4.1791e-05,
              3.8167e-05,
              3.7792e-05,
              4.0292e-05,
              3.25e-05,
              3.0208e-05,
              4.6333e-05,
              3.2833e-05,
              2.9875e-05,
              2.7791e-05,
              2.7e-05,
              3.0166e-05,
              3.4041e-05,
              2.8583e-05
            ],
            "raw_process_cpu_seconds": [
              4.8e-05,
              4.1e-05,
              4e-05,
              4.3e-05,
              6.5e-05,
              4.5e-05,
              4.9e-05,
              6e-05,
              6.7e-05,
              4.4e-05,
              5e-05,
              3.9e-05,
              4e-05,
              5.2e-05,
              5.8e-05,
              4.5e-05,
              3.9e-05,
              4.6e-05,
              4.2e-05,
              4e-05,
              4.6e-05,
              3.6e-05,
              4.3e-05,
              4.9e-05,
              3.5e-05,
              3.1e-05,
              3e-05,
              3e-05,
              3.6e-05,
              3.7e-05,
              2.9e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006766333004634362,
            "speedup_vs_optimized_serial": 0.024150696315696694
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 8.291e-06,
            "mad_seconds": 3.2080000000000005e-06,
            "median_process_cpu_seconds": 1.7e-05,
            "median_process_cpu_wall_ratio": 1.9458407653640344,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 120612.71257990591,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1.1708e-05,
              7.541e-06,
              5.5e-06,
              5.083e-06,
              7.875e-06,
              7.917e-06,
              6.458e-06,
              4.167e-06,
              9.458e-06,
              6.75e-06,
              6.167e-06,
              6.667e-06,
              1.1708e-05,
              1.3e-05,
              8.291e-06,
              6.208e-06,
              8.75e-06,
              6.375e-06,
              1.175e-05,
              3.1167e-05,
              7.125e-06,
              3.875e-06,
              1.625e-05,
              1.7875e-05,
              2.2167e-05,
              1.5333e-05,
              3.25e-06,
              1.2625e-05,
              1.3667e-05,
              1.6e-05,
              1.1e-05
            ],
            "raw_process_cpu_seconds": [
              2.1e-05,
              1.6e-05,
              1e-05,
              9e-06,
              1.5e-05,
              1.5e-05,
              1.3e-05,
              1.1e-05,
              1.8e-05,
              1.3e-05,
              1.2e-05,
              1.3e-05,
              2.2e-05,
              2.6e-05,
              1.7e-05,
              1.3e-05,
              1.8e-05,
              1.5e-05,
              2.5e-05,
              4.3e-05,
              1.1e-05,
              1e-05,
              3.4e-05,
              3.7e-05,
              3.9e-05,
              3.1e-05,
              6e-06,
              2.3e-05,
              2.8e-05,
              3.1e-05,
              2.1e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005420249995950144,
            "speedup_vs_optimized_serial": 0.11566759136412977
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 4.5167e-05,
            "mad_seconds": 5.7920000000000015e-06,
            "median_process_cpu_seconds": 6.9e-05,
            "median_process_cpu_wall_ratio": 1.4984126984126982,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 22140.05800695198,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.8834e-05,
              5.0833e-05,
              5.0834e-05,
              4.5167e-05,
              3.8625e-05,
              4.1583e-05,
              6.2334e-05,
              6.6958e-05,
              4.3e-05,
              4.1875e-05,
              5.5833e-05,
              3.8416e-05,
              5.2625e-05,
              4.9875e-05,
              3.9375e-05,
              3.9375e-05,
              4.3208e-05,
              3.6167e-05,
              4.1541e-05,
              5.2166e-05,
              4.4667e-05,
              5.0334e-05,
              7.0042e-05,
              6.3459e-05,
              3.6959e-05,
              4.2542e-05,
              7.3375e-05,
              4.3167e-05,
              3.7875e-05,
              4.8458e-05,
              5.4333e-05
            ],
            "raw_process_cpu_seconds": [
              8.9e-05,
              7.1e-05,
              8.6e-05,
              6.7e-05,
              6.3e-05,
              6.1e-05,
              9.6e-05,
              9.8e-05,
              6.2e-05,
              6.7e-05,
              8.7e-05,
              5.5e-05,
              7.6e-05,
              7.7e-05,
              6.6e-05,
              5.9e-05,
              6.1e-05,
              5.1e-05,
              6.6e-05,
              8.3e-05,
              6.7e-05,
              7.1e-05,
              0.000107,
              9.5e-05,
              5.4e-05,
              6.8e-05,
              0.000116,
              7.2e-05,
              5.6e-05,
              6.9e-05,
              7.9e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006540041998960078,
            "speedup_vs_optimized_serial": 0.021232315628666946
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 2.4e-05,
            "mad_seconds": 7.374999999999996e-06,
            "median_process_cpu_seconds": 6.7e-05,
            "median_process_cpu_wall_ratio": 2.8656716417910446,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 41666.666666666664,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1.5875e-05,
              7.2708e-05,
              2.5542e-05,
              2.6625e-05,
              2.0916e-05,
              1.6834e-05,
              1.5625e-05,
              1.5875e-05,
              2.709e-06,
              1.5666e-05,
              3.15e-05,
              2.8959e-05,
              3.35e-05,
              5.9417e-05,
              3.0083e-05,
              3.1375e-05,
              2.4625e-05,
              1.3125e-05,
              1.7583e-05,
              2.1166e-05,
              2.275e-05,
              1.3584e-05,
              4.1375e-05,
              1.9083e-05,
              3.6167e-05,
              2.7709e-05,
              1.0459e-05,
              2.4833e-05,
              2.0875e-05,
              4.0167e-05,
              2.4e-05
            ],
            "raw_process_cpu_seconds": [
              6.5e-05,
              0.000191,
              6.9e-05,
              6.8e-05,
              6e-05,
              5.5e-05,
              5.9e-05,
              4.8e-05,
              6e-06,
              5.7e-05,
              7.5e-05,
              7e-05,
              9.6e-05,
              9.8e-05,
              0.000103,
              8.2e-05,
              8.1e-05,
              4.2e-05,
              6.5e-05,
              6.7e-05,
              6.4e-05,
              5.1e-05,
              0.000114,
              5.2e-05,
              9.9e-05,
              8.2e-05,
              3.8e-05,
              6.8e-05,
              6.5e-05,
              0.000108,
              6.6e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005709999997634441,
            "speedup_vs_optimized_serial": 0.03995833333333333
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.9792e-05,
            "mad_seconds": 9.291999999999992e-06,
            "median_process_cpu_seconds": 0.000164,
            "median_process_cpu_wall_ratio": 1.960072595281307,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 12532.58472027271,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              9.2125e-05,
              7.2709e-05,
              6.7417e-05,
              8.8042e-05,
              9.8709e-05,
              9.4125e-05,
              6.8541e-05,
              7.6625e-05,
              7.2792e-05,
              7.7792e-05,
              8.5417e-05,
              7.9792e-05,
              6.8875e-05,
              8.6791e-05,
              9.3542e-05,
              6.7375e-05,
              8.2417e-05,
              0.000117125,
              8.4375e-05,
              8.6833e-05,
              8.6958e-05,
              0.000103917,
              6.6792e-05,
              6.425e-05,
              7.05e-05,
              6.475e-05,
              8.3791e-05,
              6.2292e-05,
              9.1833e-05,
              7.525e-05,
              7.225e-05
            ],
            "raw_process_cpu_seconds": [
              0.000165,
              0.000155,
              0.000144,
              0.000164,
              0.000239,
              0.000185,
              0.000145,
              0.000177,
              0.000133,
              0.000164,
              0.000168,
              0.000145,
              0.000135,
              0.000166,
              0.000173,
              0.000131,
              0.000166,
              0.000233,
              0.000168,
              0.000171,
              0.000167,
              0.000198,
              0.000126,
              0.000123,
              0.00012,
              0.000136,
              0.000162,
              0.00011,
              0.000185,
              0.000143,
              0.000144
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.0075130830009584315,
            "speedup_vs_optimized_serial": 0.012018748746741529
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "wide",
      "kernel": "exploratory-before-token-borrow",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:18:52Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "089073b5229a3f8a480c178d71c21e06f2d0a2704073e495cabc8abf5223943d"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "ac7a04f65df59b8634de88ef46b3bdd49e850b48c723d4c5b9ce4f0cc6b9e300"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.18021325,
            "mad_seconds": 0.0007421249999999824,
            "median_process_cpu_seconds": 0.179862,
            "median_process_cpu_wall_ratio": 0.9992371191049984,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 355.1348194430765,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.182969584,
              0.182498708,
              0.179471125,
              0.185364375,
              0.179851167,
              0.179956667,
              0.18021325
            ],
            "raw_process_cpu_seconds": [
              0.18283,
              0.182448,
              0.179302,
              0.185204,
              0.179738,
              0.179862,
              0.179724
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.3704980830007116
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.181763542,
            "mad_seconds": 0.0004804999999999948,
            "median_process_cpu_seconds": 0.181321,
            "median_process_cpu_wall_ratio": 0.99884280171635,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 352.1058144872639,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.18151525,
              0.183215917,
              0.180233875,
              0.18125675,
              0.182244042,
              0.181763542,
              0.182230959
            ],
            "raw_process_cpu_seconds": [
              0.181321,
              0.183062,
              0.180051,
              0.181047,
              0.181952,
              0.181288,
              0.182002
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.384824792003201,
            "speedup_vs_original_serial": 0.9914708308226079
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.18156975,
            "mad_seconds": 0.002409458000000003,
            "median_process_cpu_seconds": 0.179861,
            "median_process_cpu_wall_ratio": 0.9905890160668283,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 352.4816220763646,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.179701042,
              0.179160292,
              0.17971975,
              0.194885875,
              0.186629042,
              0.199639416,
              0.18156975
            ],
            "raw_process_cpu_seconds": [
              0.179368,
              0.178851,
              0.178944,
              0.188924,
              0.181129,
              0.1876,
              0.179861
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 3.143445540998073,
            "speedup_vs_optimized_serial": 1.0010673143516473,
            "speedup_vs_original_serial": 0.9925290418695845
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.185014042,
            "mad_seconds": 0.001749582999999999,
            "median_process_cpu_seconds": 0.183707,
            "median_process_cpu_wall_ratio": 0.9987915878441003,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 345.91968970657916,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.18691375,
              0.18716375,
              0.183264459,
              0.185014042,
              0.183272,
              0.185048125,
              0.181173084
            ],
            "raw_process_cpu_seconds": [
              0.183707,
              0.186901,
              0.183043,
              0.184817,
              0.183142,
              0.184983,
              0.180939
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.4016249170017545,
            "speedup_vs_optimized_serial": 0.982431063259512,
            "speedup_vs_original_serial": 0.9740517425158465
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.094001875,
            "mad_seconds": 0.00028200000000000447,
            "median_process_cpu_seconds": 0.187421,
            "median_process_cpu_wall_ratio": 1.991872496906856,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 680.837483294881,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.093719875,
              0.094230166,
              0.094001875,
              0.093827292,
              0.092308584,
              0.094633292,
              0.095642166
            ],
            "raw_process_cpu_seconds": [
              0.18667,
              0.188134,
              0.186651,
              0.187421,
              0.183911,
              0.188127,
              0.190507
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.05508916600229,
            "speedup_vs_optimized_serial": 1.9336161326569283,
            "speedup_vs_original_serial": 1.917123993537363
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.094377167,
            "mad_seconds": 0.0005990419999999941,
            "median_process_cpu_seconds": 0.188441,
            "median_process_cpu_wall_ratio": 1.9938349678599465,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 678.130124418759,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.094377167,
              0.094133167,
              0.093778125,
              0.095809,
              0.095003,
              0.09475425,
              0.093565334
            ],
            "raw_process_cpu_seconds": [
              0.188441,
              0.187686,
              0.18725,
              0.190952,
              0.189234,
              0.189201,
              0.185992
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.6805167079946841,
            "speedup_vs_optimized_serial": 1.925927083613349,
            "speedup_vs_original_serial": 1.9095005256938895
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.050122542,
            "mad_seconds": 0.00043120800000000237,
            "median_process_cpu_seconds": 0.19937,
            "median_process_cpu_wall_ratio": 3.97518274082839,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1276.8705944722437,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.050122542,
              0.05016025,
              0.050653125,
              0.049958458,
              0.049239792,
              0.05055375,
              0.049019541
            ],
            "raw_process_cpu_seconds": [
              0.19937,
              0.19937,
              0.201145,
              0.198594,
              0.196248,
              0.201043,
              0.194158
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.5279741660051513,
            "speedup_vs_optimized_serial": 3.626383155108135,
            "speedup_vs_original_serial": 3.5954531196761725
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.050326,
            "mad_seconds": 0.00025983299999999404,
            "median_process_cpu_seconds": 0.199717,
            "median_process_cpu_wall_ratio": 3.964667340755267,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1271.7084608353534,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.050585833,
              0.051692625,
              0.049525708,
              0.050326,
              0.050168833,
              0.050063708,
              0.050422958
            ],
            "raw_process_cpu_seconds": [
              0.200556,
              0.204941,
              0.197247,
              0.200353,
              0.198263,
              0.199637,
              0.199717
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.3257019160009804,
            "speedup_vs_optimized_serial": 3.611722409887533,
            "speedup_vs_original_serial": 3.580917418431824
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.18227175,
            "mad_seconds": 0.0008644999999999903,
            "median_process_cpu_seconds": 0.18219,
            "median_process_cpu_wall_ratio": 0.9995514938546427,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 351.12407709916647,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.18227175,
              0.18313625,
              0.183032167,
              0.181240375,
              0.181564042,
              0.183255875,
              0.180593625
            ],
            "raw_process_cpu_seconds": [
              0.18219,
              0.183076,
              0.183017,
              0.181159,
              0.181546,
              0.183171,
              0.180447
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.3802965419963584
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.184500666,
            "mad_seconds": 0.0030561250000000206,
            "median_process_cpu_seconds": 0.184406,
            "median_process_cpu_wall_ratio": 0.9992904535623701,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 346.88221667449153,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.181218583,
              0.19191,
              0.181444541,
              0.184500666,
              0.185453584,
              0.179715458,
              0.184609709
            ],
            "raw_process_cpu_seconds": [
              0.18109,
              0.188711,
              0.1814,
              0.184462,
              0.185399,
              0.179483,
              0.184406
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 3.132321375000174,
            "speedup_vs_optimized_serial": 0.9879191980802932
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.183868083,
            "mad_seconds": 0.000632417000000024,
            "median_process_cpu_seconds": 0.183789,
            "median_process_cpu_wall_ratio": 0.9993939476294657,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 348.0756363789359,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.183868083,
              0.181047291,
              0.182164417,
              0.1839775,
              0.1845005,
              0.183963791,
              0.180478916
            ],
            "raw_process_cpu_seconds": [
              0.183859,
              0.18103,
              0.182113,
              0.183866,
              0.184359,
              0.183789,
              0.180311
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.3826776659989264,
            "speedup_vs_optimized_serial": 0.9913180527367549
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.0933955,
            "mad_seconds": 0.0009832910000000111,
            "median_process_cpu_seconds": 0.185821,
            "median_process_cpu_wall_ratio": 1.983113734607221,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 685.2578550358421,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.0933955,
              0.092084375,
              0.093276042,
              0.092412209,
              0.093774709,
              0.110613584,
              0.096641584
            ],
            "raw_process_cpu_seconds": [
              0.186271,
              0.183297,
              0.184977,
              0.184079,
              0.185821,
              0.202673,
              0.186597
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.0526143749957555,
            "speedup_vs_optimized_serial": 1.9516116943535824
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.093418167,
            "mad_seconds": 0.0007926669999999969,
            "median_process_cpu_seconds": 0.185091,
            "median_process_cpu_wall_ratio": 1.9855737560327555,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 685.0915839528301,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.090872875,
              0.09070275,
              0.093418167,
              0.095774208,
              0.093938083,
              0.0926255,
              0.093998792
            ],
            "raw_process_cpu_seconds": [
              0.180992,
              0.180097,
              0.185091,
              0.190765,
              0.186459,
              0.183997,
              0.186546
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.6632268329994986,
            "speedup_vs_optimized_serial": 1.9511381549586604
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.049588208,
            "mad_seconds": 0.0004250829999999997,
            "median_process_cpu_seconds": 0.197515,
            "median_process_cpu_wall_ratio": 3.975642757649157,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1290.6294173808417,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.049163125,
              0.049581791,
              0.049588208,
              0.050179875,
              0.049892667,
              0.051037167,
              0.04878775
            ],
            "raw_process_cpu_seconds": [
              0.195779,
              0.197515,
              0.197145,
              0.199322,
              0.198068,
              0.20266,
              0.194073
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.5325945420045173,
            "speedup_vs_optimized_serial": 3.675707539179476
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.04981275,
            "mad_seconds": 0.0004959999999999964,
            "median_process_cpu_seconds": 0.197879,
            "median_process_cpu_wall_ratio": 3.9769231543050294,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1284.811619515084,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.0497035,
              0.05030875,
              0.049696125,
              0.050828833,
              0.048538,
              0.051209791,
              0.04981275
            ],
            "raw_process_cpu_seconds": [
              0.197667,
              0.200444,
              0.197055,
              0.202097,
              0.193283,
              0.203801,
              0.197879
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.325552374997642,
            "speedup_vs_optimized_serial": 3.6591384735835706
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "final-batch",
      "kernel": "final-borrowed-stop-token",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:21:42Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.056662417,
            "mad_seconds": 0.00022241600000000306,
            "median_process_cpu_seconds": 0.056584,
            "median_process_cpu_wall_ratio": 0.9985139473219936,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4517.985881188231,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056534583,
              0.056212875,
              0.056662417,
              0.056834459,
              0.05726875,
              0.05697875,
              0.056884833,
              0.056030167,
              0.056651625
            ],
            "raw_process_cpu_seconds": [
              0.056421,
              0.056165,
              0.056584,
              0.05675,
              0.05719,
              0.056891,
              0.056777,
              0.055962,
              0.056529
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8789990420045797
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.058566,
            "mad_seconds": 0.0005084169999999971,
            "median_process_cpu_seconds": 0.058453,
            "median_process_cpu_wall_ratio": 0.9985088475048048,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4371.136837072704,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.058941833,
              0.059074417,
              0.058566,
              0.059932541,
              0.060356,
              0.058170125,
              0.057523334,
              0.058015042,
              0.058453416
            ],
            "raw_process_cpu_seconds": [
              0.058882,
              0.05899,
              0.058453,
              0.059823,
              0.060266,
              0.058091,
              0.057435,
              0.057904,
              0.058368
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0920652079948923,
            "speedup_vs_original_serial": 0.9674967899463852
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057864625,
            "mad_seconds": 0.00011312500000000558,
            "median_process_cpu_seconds": 0.057776,
            "median_process_cpu_wall_ratio": 0.9986565482081435,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4424.1192265568125,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057799208,
              0.0577515,
              0.057864625,
              0.057169541,
              0.056588417,
              0.057935834,
              0.058676208,
              0.057875875,
              0.058084584
            ],
            "raw_process_cpu_seconds": [
              0.057743,
              0.057706,
              0.057776,
              0.057053,
              0.056519,
              0.057858,
              0.058587,
              0.057815,
              0.057989
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.1262474999966798,
            "speedup_vs_optimized_serial": 1.0121209633692432,
            "speedup_vs_original_serial": 0.9792237830971858
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057819,
            "mad_seconds": 0.0002354999999999996,
            "median_process_cpu_seconds": 0.057744,
            "median_process_cpu_wall_ratio": 0.9984337454102911,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4427.610301112091,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.059215542,
              0.058248166,
              0.0580545,
              0.057984458,
              0.057741375,
              0.05678725,
              0.056716833,
              0.057819,
              0.057655208
            ],
            "raw_process_cpu_seconds": [
              0.059153,
              0.058127,
              0.05798,
              0.057892,
              0.057599,
              0.056687,
              0.056628,
              0.057744,
              0.057571
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8944127080030739,
            "speedup_vs_optimized_serial": 1.012919628495823,
            "speedup_vs_original_serial": 0.979996489043394
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028978084,
            "mad_seconds": 7.304099999999938e-05,
            "median_process_cpu_seconds": 0.057581,
            "median_process_cpu_wall_ratio": 1.985912765856744,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8834.262472287677,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.028685708,
              0.028839208,
              0.029352,
              0.028915375,
              0.028978084,
              0.029051125,
              0.029030416,
              0.028911958,
              0.029198333
            ],
            "raw_process_cpu_seconds": [
              0.057048,
              0.057234,
              0.058266,
              0.057127,
              0.05755,
              0.057693,
              0.057719,
              0.057581,
              0.057899
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7233359579986427,
            "speedup_vs_optimized_serial": 2.0210445935625003,
            "speedup_vs_original_serial": 1.9553541566102162
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028963542,
            "mad_seconds": 0.00013087499999999905,
            "median_process_cpu_seconds": 0.057522,
            "median_process_cpu_wall_ratio": 1.983428944594225,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8838.69797416352,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.029542,
              0.028741417,
              0.029144833,
              0.028853834,
              0.029116625,
              0.028963542,
              0.029001291,
              0.028934583,
              0.028832667
            ],
            "raw_process_cpu_seconds": [
              0.058231,
              0.057165,
              0.057659,
              0.057457,
              0.058035,
              0.057256,
              0.057522,
              0.057697,
              0.057167
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.6107667499964009,
            "speedup_vs_optimized_serial": 2.0220593185736746,
            "speedup_vs_original_serial": 1.9563358998012053
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014409292,
            "mad_seconds": 5.93750000000004e-05,
            "median_process_cpu_seconds": 0.057417,
            "median_process_cpu_wall_ratio": 3.9786459069690645,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17766.313570437742,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014626958,
              0.014595917,
              0.014636916,
              0.014664542,
              0.014349917,
              0.014368125,
              0.014407917,
              0.014409292,
              0.014389167
            ],
            "raw_process_cpu_seconds": [
              0.057707,
              0.057677,
              0.057831,
              0.057876,
              0.057274,
              0.05732,
              0.057324,
              0.057417,
              0.057296
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.5222969580063364,
            "speedup_vs_optimized_serial": 4.064460627211941,
            "speedup_vs_original_serial": 3.932352609691024
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014980625,
            "mad_seconds": 9.204199999999878e-05,
            "median_process_cpu_seconds": 0.058375,
            "median_process_cpu_wall_ratio": 3.8896306528302094,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17088.739622011766,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014985417,
              0.015039667,
              0.014888583,
              0.014767083,
              0.015082666,
              0.014980625,
              0.01525425,
              0.014960084,
              0.014673167
            ],
            "raw_process_cpu_seconds": [
              0.058243,
              0.058375,
              0.058375,
              0.058085,
              0.058666,
              0.058496,
              0.058378,
              0.058164,
              0.057775
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.4685773329983931,
            "speedup_vs_optimized_serial": 3.9094497058700823,
            "speedup_vs_original_serial": 3.782380040886145
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.058394292,
            "mad_seconds": 0.0002969589999999994,
            "median_process_cpu_seconds": 0.058321,
            "median_process_cpu_wall_ratio": 0.9987448773246536,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4383.990133830203,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.059471875,
              0.058097333,
              0.057755167,
              0.058541166,
              0.058394292,
              0.057681917,
              0.058193833,
              0.058431125,
              0.059397209
            ],
            "raw_process_cpu_seconds": [
              0.059374,
              0.058031,
              0.057685,
              0.058423,
              0.058321,
              0.05762,
              0.058119,
              0.058338,
              0.059345
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9063728750043083
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.05776375,
            "mad_seconds": 0.00016975000000000323,
            "median_process_cpu_seconds": 0.057701,
            "median_process_cpu_wall_ratio": 0.9986424158036724,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4431.845231654801,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057974625,
              0.05906375,
              0.057594,
              0.057731125,
              0.057455,
              0.058006333,
              0.057734084,
              0.057789833,
              0.05776375
            ],
            "raw_process_cpu_seconds": [
              0.05781,
              0.058936,
              0.05753,
              0.057624,
              0.057377,
              0.057928,
              0.057645,
              0.057713,
              0.057701
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.1384263340005418,
            "speedup_vs_optimized_serial": 1.0109158771721019
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.058085084,
            "mad_seconds": 0.00032695799999999525,
            "median_process_cpu_seconds": 0.057971,
            "median_process_cpu_wall_ratio": 0.9986499574488057,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4407.327705680859,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057478708,
              0.0569945,
              0.057441875,
              0.058412042,
              0.058085084,
              0.058014834,
              0.058383916,
              0.058268667,
              0.058763333
            ],
            "raw_process_cpu_seconds": [
              0.057434,
              0.056875,
              0.057385,
              0.05834,
              0.057971,
              0.057932,
              0.058256,
              0.058214,
              0.058684
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.89836420899519,
            "speedup_vs_optimized_serial": 1.0053233632235083
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028858917,
            "mad_seconds": 0.00013187500000000005,
            "median_process_cpu_seconds": 0.057523,
            "median_process_cpu_wall_ratio": 1.9901084268594256,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8870.741753753267,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.02904775,
              0.028954792,
              0.028727042,
              0.028643875,
              0.028665333,
              0.0289135,
              0.028858917,
              0.029023292,
              0.028807667
            ],
            "raw_process_cpu_seconds": [
              0.057702,
              0.057523,
              0.057315,
              0.0572,
              0.057249,
              0.057541,
              0.057629,
              0.057744,
              0.057311
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7219405840005493,
            "speedup_vs_optimized_serial": 2.0234401727549236
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.029109583,
            "mad_seconds": 0.00018112500000000073,
            "median_process_cpu_seconds": 0.057745,
            "median_process_cpu_wall_ratio": 1.9892761775391972,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8794.354766263741,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.029226,
              0.029091125,
              0.028928458,
              0.029324084,
              0.029401833,
              0.029109583,
              0.02867975,
              0.028919917,
              0.029172
            ],
            "raw_process_cpu_seconds": [
              0.057722,
              0.058043,
              0.057722,
              0.058174,
              0.058051,
              0.057907,
              0.057197,
              0.057537,
              0.057745
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.6082527079997817,
            "speedup_vs_optimized_serial": 2.006016094424987
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014634292,
            "mad_seconds": 0.00015687499999999903,
            "median_process_cpu_seconds": 0.057625,
            "median_process_cpu_wall_ratio": 3.9376691403998225,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17493.159218088582,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.015031958,
              0.014747,
              0.014791167,
              0.014667167,
              0.01442025,
              0.01444075,
              0.01456,
              0.014467208,
              0.014634292
            ],
            "raw_process_cpu_seconds": [
              0.058081,
              0.058043,
              0.058037,
              0.057729,
              0.057346,
              0.057447,
              0.057513,
              0.057534,
              0.057625
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.5205321670000558,
            "speedup_vs_optimized_serial": 3.990236903842017
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.01463775,
            "mad_seconds": 5.766599999999962e-05,
            "median_process_cpu_seconds": 0.05776,
            "median_process_cpu_wall_ratio": 3.949693934335002,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17489.026660518182,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.01456025,
              0.01456475,
              0.014600625,
              0.01463775,
              0.014830167,
              0.014658,
              0.014714125,
              0.014630542,
              0.014695416
            ],
            "raw_process_cpu_seconds": [
              0.057625,
              0.057603,
              0.057668,
              0.05776,
              0.057996,
              0.0581,
              0.057838,
              0.057712,
              0.058183
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.4643039169968688,
            "speedup_vs_optimized_serial": 3.989294256289389
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "final-tiny",
      "kernel": "final-borrowed-stop-token",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:21:54Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.08e-07,
            "mad_seconds": 9.999999999999692e-10,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.4124293785310733,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1412429.3785310735,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1.833e-06,
              7.92e-07,
              7.5e-07,
              7.08e-07,
              6.67e-07,
              7.5e-07,
              7.08e-07,
              7.09e-07,
              7.08e-07,
              7.08e-07,
              7.08e-07,
              7.5e-07,
              7.08e-07,
              7.08e-07,
              6.67e-07,
              6.67e-07,
              7.09e-07,
              7.08e-07,
              7.09e-07,
              7.08e-07,
              7.5e-07,
              6.66e-07,
              7.5e-07,
              7.09e-07,
              7.08e-07,
              7.5e-07,
              6.66e-07,
              7.92e-07,
              7.09e-07,
              7.08e-07,
              7.08e-07
            ],
            "raw_process_cpu_seconds": [
              3e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005090416998427827
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.92e-07,
            "mad_seconds": 9.999999999999692e-10,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.2626262626262625,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1262626.2626262626,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.583e-06,
              2.083e-06,
              9.58e-07,
              8.75e-07,
              8.34e-07,
              8.33e-07,
              8.33e-07,
              7.92e-07,
              7.91e-07,
              8.33e-07,
              8.34e-07,
              7.91e-07,
              7.5e-07,
              7.92e-07,
              7.92e-07,
              7.91e-07,
              7.92e-07,
              7.92e-07,
              7.92e-07,
              7.91e-07,
              7.92e-07,
              7.92e-07,
              8.33e-07,
              7.92e-07,
              7.92e-07,
              7.92e-07,
              7.91e-07,
              7.92e-07,
              7.5e-07,
              8.33e-07,
              7.92e-07
            ],
            "raw_process_cpu_seconds": [
              6e-06,
              3e-06,
              1e-06,
              0,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              0,
              0,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.004778624999744352,
            "speedup_vs_original_serial": 0.893939393939394
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.917e-06,
            "mad_seconds": 1.8329999999999992e-06,
            "median_process_cpu_seconds": 1.1e-05,
            "median_process_cpu_wall_ratio": 1.3602798289933928,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 126310.47113805734,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              5.291e-06,
              9.416e-06,
              1.4e-05,
              1.1834e-05,
              7.917e-06,
              5e-06,
              9.5e-06,
              9.458e-06,
              1.9041e-05,
              9.25e-06,
              8.375e-06,
              7.917e-06,
              6.625e-06,
              4.916e-06,
              1.1125e-05,
              9.75e-06,
              6.667e-06,
              6.875e-06,
              8.084e-06,
              4.916e-06,
              8.542e-06,
              4.167e-06,
              4.666e-06,
              4.833e-06,
              6.042e-06,
              4.416e-06,
              4.667e-06,
              1.0292e-05,
              8.583e-06,
              7.167e-06,
              9.208e-06
            ],
            "raw_process_cpu_seconds": [
              1e-05,
              1.3e-05,
              1.8e-05,
              1.6e-05,
              1e-05,
              7e-06,
              1.2e-05,
              1.1e-05,
              2.2e-05,
              1.1e-05,
              1.2e-05,
              1.2e-05,
              1.1e-05,
              7e-06,
              1.3e-05,
              1.3e-05,
              8e-06,
              8e-06,
              1.2e-05,
              9e-06,
              1.1e-05,
              6e-06,
              8e-06,
              8e-06,
              9e-06,
              6e-06,
              8e-06,
              1.4e-05,
              1.1e-05,
              1e-05,
              1.2e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.004873792000580579,
            "speedup_vs_optimized_serial": 0.10003789314134141,
            "speedup_vs_original_serial": 0.0894278135657446
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 2.9959e-05,
            "mad_seconds": 2.8340000000000003e-06,
            "median_process_cpu_seconds": 3.5e-05,
            "median_process_cpu_wall_ratio": 1.1936922786957342,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 33378.95123335225,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.6625e-05,
              5.1166e-05,
              3.2083e-05,
              3.625e-05,
              2.6292e-05,
              2.6667e-05,
              2.7125e-05,
              2.8375e-05,
              2.9417e-05,
              3.0125e-05,
              2.5083e-05,
              2.4625e-05,
              3.1833e-05,
              3.4542e-05,
              3.2334e-05,
              2.7625e-05,
              2.7125e-05,
              3.1834e-05,
              3.2542e-05,
              2.7041e-05,
              2.6667e-05,
              3.7333e-05,
              2.9e-05,
              2.9959e-05,
              2.8917e-05,
              2.9166e-05,
              3.7083e-05,
              4.5792e-05,
              3.0917e-05,
              3.0375e-05,
              2.6417e-05
            ],
            "raw_process_cpu_seconds": [
              5.6e-05,
              5.6e-05,
              3.9e-05,
              4.3e-05,
              3.4e-05,
              3.1e-05,
              3.1e-05,
              3.4e-05,
              3.5e-05,
              3.4e-05,
              3e-05,
              3e-05,
              3.9e-05,
              4.1e-05,
              3.8e-05,
              3.3e-05,
              3.3e-05,
              3.8e-05,
              3.9e-05,
              3.1e-05,
              3.4e-05,
              4.9e-05,
              3.2e-05,
              3.6e-05,
              3.3e-05,
              3.5e-05,
              4.4e-05,
              5.3e-05,
              3.7e-05,
              3.4e-05,
              3.1e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005805875000078231,
            "speedup_vs_optimized_serial": 0.02643612937681498,
            "speedup_vs_original_serial": 0.023632297473213392
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.5e-06,
            "mad_seconds": 1.833e-06,
            "median_process_cpu_seconds": 1.5e-05,
            "median_process_cpu_wall_ratio": 2,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 133333.33333333334,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1.1583e-05,
              6e-06,
              3.542e-06,
              8.084e-06,
              6.5e-06,
              8.791e-06,
              7.583e-06,
              5.667e-06,
              7.375e-06,
              1.0667e-05,
              9e-06,
              5.25e-06,
              2.542e-06,
              8.958e-06,
              7.458e-06,
              2.875e-06,
              9.542e-06,
              1e-05,
              9.167e-06,
              1.5834e-05,
              1.0666e-05,
              1.15e-05,
              1.5e-05,
              7.5e-06,
              6.125e-06,
              6.166e-06,
              5.667e-06,
              5.375e-06,
              8.667e-06,
              5.5e-06,
              2.833e-06
            ],
            "raw_process_cpu_seconds": [
              2.1e-05,
              1.2e-05,
              9e-06,
              1.5e-05,
              1.1e-05,
              1.8e-05,
              1.4e-05,
              1.1e-05,
              1.5e-05,
              1.9e-05,
              1.6e-05,
              1e-05,
              6e-06,
              1.8e-05,
              1.5e-05,
              7e-06,
              2e-05,
              1.9e-05,
              2.2e-05,
              2.2e-05,
              2.6e-05,
              2.3e-05,
              3.2e-05,
              1.6e-05,
              1e-05,
              1.2e-05,
              1.1e-05,
              1.1e-05,
              1.6e-05,
              1e-05,
              6e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005363292002584785,
            "speedup_vs_optimized_serial": 0.1056,
            "speedup_vs_original_serial": 0.0944
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 4.8458e-05,
            "mad_seconds": 5.334e-06,
            "median_process_cpu_seconds": 7.3e-05,
            "median_process_cpu_wall_ratio": 1.4824281150159744,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 20636.427421684755,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              5.7667e-05,
              5.4833e-05,
              4.8042e-05,
              3.9666e-05,
              4.0583e-05,
              5.0375e-05,
              4.1958e-05,
              4.5958e-05,
              7.2792e-05,
              5.1792e-05,
              5.0125e-05,
              4.3125e-05,
              3.8916e-05,
              4.7375e-05,
              3.6417e-05,
              4.5416e-05,
              5.6667e-05,
              4.8458e-05,
              3.9125e-05,
              5.3792e-05,
              5.1583e-05,
              4.1542e-05,
              4.9333e-05,
              4.6375e-05,
              5.9042e-05,
              5.1333e-05,
              4.8083e-05,
              5.4125e-05,
              4.7833e-05,
              5.9792e-05,
              5.525e-05
            ],
            "raw_process_cpu_seconds": [
              8e-05,
              7.8e-05,
              7.6e-05,
              6.1e-05,
              6.6e-05,
              6.9e-05,
              6.1e-05,
              7.3e-05,
              0.000109,
              7.3e-05,
              7.9e-05,
              5.8e-05,
              5.4e-05,
              7.2e-05,
              5.6e-05,
              7.7e-05,
              8.5e-05,
              6.5e-05,
              5.8e-05,
              8.3e-05,
              7.2e-05,
              5.8e-05,
              6.7e-05,
              6.1e-05,
              8.7e-05,
              7.8e-05,
              7.6e-05,
              8.6e-05,
              6.6e-05,
              8.5e-05,
              8.6e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006553040999278892,
            "speedup_vs_optimized_serial": 0.01634405051797433,
            "speedup_vs_original_serial": 0.01461059061455281
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 1.9875e-05,
            "mad_seconds": 8.458e-06,
            "median_process_cpu_seconds": 6e-05,
            "median_process_cpu_wall_ratio": 3.015362047186341,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 50314.46540880504,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              2.7125e-05,
              2.0875e-05,
              1.9416e-05,
              1.9875e-05,
              6.542e-06,
              3.0542e-05,
              2.6458e-05,
              3.6041e-05,
              1.1417e-05,
              1.7083e-05,
              2.6791e-05,
              1.1083e-05,
              2.7417e-05,
              1.6458e-05,
              3.9416e-05,
              4.7333e-05,
              3.125e-05,
              9.959e-06,
              2.1209e-05,
              1.5875e-05,
              6.625e-06,
              3.0958e-05,
              2.2542e-05,
              4.416e-06,
              1.3875e-05,
              4.8208e-05,
              2.4541e-05,
              1.1084e-05,
              1.1416e-05,
              1.3875e-05,
              7.625e-06
            ],
            "raw_process_cpu_seconds": [
              9.2e-05,
              5.9e-05,
              5.1e-05,
              6.3e-05,
              3e-05,
              9.1e-05,
              6.7e-05,
              9.7e-05,
              2.8e-05,
              5.2e-05,
              6.4e-05,
              3.9e-05,
              9.2e-05,
              6e-05,
              0.000123,
              0.00011,
              7.9e-05,
              2.7e-05,
              6.8e-05,
              4.8e-05,
              2.5e-05,
              0.000106,
              6.2e-05,
              2e-05,
              3.8e-05,
              0.000124,
              7.4e-05,
              2.8e-05,
              3.8e-05,
              4.7e-05,
              2.2e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.0060092919957241975,
            "speedup_vs_optimized_serial": 0.03984905660377359,
            "speedup_vs_original_serial": 0.035622641509433964
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.8417e-05,
            "mad_seconds": 8.583000000000009e-06,
            "median_process_cpu_seconds": 0.000157,
            "median_process_cpu_wall_ratio": 2.039730400851366,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 12752.336865730644,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              9.8875e-05,
              7.7375e-05,
              6.4125e-05,
              6.2584e-05,
              6.0916e-05,
              7.85e-05,
              8.45e-05,
              0.000104417,
              8.6667e-05,
              8.8459e-05,
              8.85e-05,
              6.65e-05,
              6.4083e-05,
              7.9875e-05,
              7.675e-05,
              5.75e-05,
              8.35e-05,
              8.2083e-05,
              9.7041e-05,
              7.3375e-05,
              8.2292e-05,
              9.0208e-05,
              7.6667e-05,
              6.9834e-05,
              9.9583e-05,
              9.3584e-05,
              6.0083e-05,
              7.3125e-05,
              7.0125e-05,
              7.5875e-05,
              7.8417e-05
            ],
            "raw_process_cpu_seconds": [
              0.0002,
              0.000173,
              0.000111,
              0.000121,
              0.000109,
              0.000162,
              0.00017,
              0.000217,
              0.000178,
              0.000205,
              0.000163,
              0.000137,
              0.000135,
              0.000135,
              0.000169,
              0.000108,
              0.000173,
              0.000151,
              0.000187,
              0.00015,
              0.000155,
              0.000184,
              0.000154,
              0.000146,
              0.000198,
              0.000206,
              0.000121,
              0.000151,
              0.000151,
              0.000182,
              0.000157
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.007696749999013264,
            "speedup_vs_optimized_serial": 0.01009985079765867,
            "speedup_vs_original_serial": 0.009028654500937297
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 9.17e-07,
            "mad_seconds": 4.199999999999998e-08,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.1428571428571428,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1090512.5408942204,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1e-06,
              9.58e-07,
              9.58e-07,
              9.58e-07,
              2.25e-06,
              1.042e-06,
              9.59e-07,
              1.125e-06,
              2.208e-06,
              1.125e-06,
              9.59e-07,
              8.75e-07,
              8.75e-07,
              8.75e-07,
              9.59e-07,
              8.75e-07,
              8.75e-07,
              9.17e-07,
              9.17e-07,
              8.75e-07,
              9.16e-07,
              9.16e-07,
              9.16e-07,
              9.17e-07,
              8.75e-07,
              9.59e-07,
              9.59e-07,
              8.75e-07,
              9.16e-07,
              8.75e-07,
              9.16e-07
            ],
            "raw_process_cpu_seconds": [
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              3e-06,
              1e-06,
              1e-06,
              1e-06,
              3e-06,
              1e-06,
              2e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.004915958001220133
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 8.833e-06,
            "mad_seconds": 3.7909999999999998e-06,
            "median_process_cpu_seconds": 1.2e-05,
            "median_process_cpu_wall_ratio": 1.354751306367331,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 113211.81931393637,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              5.042e-06,
              4.458e-06,
              1.0375e-05,
              4.958e-06,
              4.875e-06,
              1.3375e-05,
              1.0167e-05,
              9.375e-06,
              7.917e-06,
              5.5e-06,
              1.3042e-05,
              8.833e-06,
              1.5667e-05,
              1.9958e-05,
              8.708e-06,
              5.958e-06,
              5.083e-06,
              5.125e-06,
              5.167e-06,
              5.125e-06,
              4.625e-06,
              4.541e-06,
              6.125e-06,
              1.0459e-05,
              2.05e-05,
              2.0958e-05,
              3.0417e-05,
              1.3875e-05,
              1.5333e-05,
              1.3708e-05,
              1.0542e-05
            ],
            "raw_process_cpu_seconds": [
              7e-06,
              6e-06,
              1.4e-05,
              7e-06,
              7e-06,
              1.8e-05,
              1.2e-05,
              1.5e-05,
              1.1e-05,
              8e-06,
              1.5e-05,
              1.3e-05,
              1.7e-05,
              2e-05,
              1.6e-05,
              8e-06,
              6e-06,
              7e-06,
              7e-06,
              7e-06,
              7e-06,
              8e-06,
              9e-06,
              1.2e-05,
              2.5e-05,
              2.5e-05,
              2.6e-05,
              1.9e-05,
              2.1e-05,
              1.8e-05,
              1.2e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005367416000808589,
            "speedup_vs_optimized_serial": 0.10381523831087965
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 3.1416e-05,
            "mad_seconds": 3.333e-06,
            "median_process_cpu_seconds": 3.8e-05,
            "median_process_cpu_wall_ratio": 1.186602870813397,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 31830.914183855362,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.1833e-05,
              3.8625e-05,
              3.5375e-05,
              3.2209e-05,
              3.575e-05,
              3.175e-05,
              3.1416e-05,
              2.9667e-05,
              3.025e-05,
              3.1417e-05,
              2.375e-05,
              2.5042e-05,
              3.7166e-05,
              3.6125e-05,
              3.4125e-05,
              3.2792e-05,
              3.0333e-05,
              2.8541e-05,
              3.5875e-05,
              3.15e-05,
              2.675e-05,
              2.4666e-05,
              3.2916e-05,
              3.125e-05,
              2.6125e-05,
              2.5125e-05,
              2.85e-05,
              2.8667e-05,
              2.8083e-05,
              2.6041e-05,
              4.2708e-05
            ],
            "raw_process_cpu_seconds": [
              4.7e-05,
              4.7e-05,
              4.4e-05,
              4e-05,
              4.2e-05,
              3.8e-05,
              4e-05,
              3.4e-05,
              4.5e-05,
              3.8e-05,
              2.8e-05,
              3e-05,
              4.3e-05,
              3.9e-05,
              3.9e-05,
              3.8e-05,
              3.8e-05,
              3.5e-05,
              4.1e-05,
              3.6e-05,
              3.4e-05,
              3e-05,
              3.6e-05,
              3.4e-05,
              3.1e-05,
              3e-05,
              3.5e-05,
              3.4e-05,
              3.4e-05,
              2.9e-05,
              5e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006231792001926806,
            "speedup_vs_optimized_serial": 0.029188948306595366
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 1.0375e-05,
            "mad_seconds": 3.1249999999999997e-06,
            "median_process_cpu_seconds": 2e-05,
            "median_process_cpu_wall_ratio": 1.8909090909090907,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 96385.5421686747,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              7.583e-06,
              2.475e-05,
              1.0375e-05,
              7.625e-06,
              7.042e-06,
              1.6375e-05,
              1.0667e-05,
              5e-06,
              8.125e-06,
              6.667e-06,
              7.958e-06,
              3.791e-06,
              5.583e-06,
              9.125e-06,
              7.25e-06,
              6.875e-06,
              2.2125e-05,
              7.666e-06,
              3.3209e-05,
              1.8625e-05,
              1.225e-05,
              1.65e-05,
              1.3792e-05,
              2.0333e-05,
              1.2e-05,
              6.625e-06,
              1.0333e-05,
              1.1875e-05,
              1.1791e-05,
              1.0667e-05,
              1.075e-05
            ],
            "raw_process_cpu_seconds": [
              1.4e-05,
              5.1e-05,
              2.2e-05,
              1.3e-05,
              1.3e-05,
              2.9e-05,
              2.1e-05,
              1e-05,
              1.5e-05,
              1.4e-05,
              1.6e-05,
              9e-06,
              1.1e-05,
              1.6e-05,
              1.2e-05,
              1.3e-05,
              3.6e-05,
              1.4e-05,
              5.5e-05,
              3.1e-05,
              2e-05,
              3.2e-05,
              2.5e-05,
              3.9e-05,
              2.5e-05,
              1.9e-05,
              1.8e-05,
              2.5e-05,
              2e-05,
              2.1e-05,
              2.1e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005499500002770219,
            "speedup_vs_optimized_serial": 0.0883855421686747
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 4.9417e-05,
            "mad_seconds": 7.041999999999998e-06,
            "median_process_cpu_seconds": 7.5e-05,
            "median_process_cpu_wall_ratio": 1.492794396279497,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 20235.951190885728,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              6.3959e-05,
              5.8083e-05,
              4.0167e-05,
              4.2375e-05,
              4.4958e-05,
              5.5208e-05,
              5.9292e-05,
              7.9958e-05,
              5.2916e-05,
              5.2917e-05,
              4.9209e-05,
              3.9833e-05,
              4.6e-05,
              4.7375e-05,
              3.9584e-05,
              3.4834e-05,
              4.9417e-05,
              5.5458e-05,
              4.0792e-05,
              4.9542e-05,
              6.7333e-05,
              5.1584e-05,
              5.1916e-05,
              6.8041e-05,
              3.5166e-05,
              4.0208e-05,
              4.5875e-05,
              5.0375e-05,
              5.7708e-05,
              3.8542e-05,
              4.5417e-05
            ],
            "raw_process_cpu_seconds": [
              0.000108,
              7.7e-05,
              5.9e-05,
              6.4e-05,
              6.8e-05,
              9e-05,
              9.1e-05,
              0.000119,
              9.3e-05,
              7.6e-05,
              7.5e-05,
              5.5e-05,
              7.6e-05,
              7.7e-05,
              5.8e-05,
              5.2e-05,
              7.7e-05,
              7.5e-05,
              5.7e-05,
              6.9e-05,
              0.000104,
              6.6e-05,
              7.5e-05,
              9.6e-05,
              5.6e-05,
              5.8e-05,
              6.4e-05,
              7.7e-05,
              7.9e-05,
              6.3e-05,
              7.3e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.0065227499944739975,
            "speedup_vs_optimized_serial": 0.018556367242042213
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 1.5792e-05,
            "mad_seconds": 7.417e-06,
            "median_process_cpu_seconds": 5.3e-05,
            "median_process_cpu_wall_ratio": 3.073502484959456,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 63323.20162107396,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              2.6625e-05,
              1.5125e-05,
              9.459e-06,
              3e-06,
              8e-06,
              2.2792e-05,
              2.8625e-05,
              3.958e-06,
              1.4709e-05,
              2.3209e-05,
              2.3083e-05,
              2.3084e-05,
              2.3166e-05,
              3.0542e-05,
              7.959e-06,
              4.125e-06,
              2.2792e-05,
              2.4709e-05,
              4.1125e-05,
              2.3209e-05,
              3.0708e-05,
              1.5792e-05,
              4.084e-06,
              7.542e-06,
              1.7334e-05,
              9.791e-06,
              1.2709e-05,
              1.5625e-05,
              8e-06,
              1.6042e-05,
              1.5292e-05
            ],
            "raw_process_cpu_seconds": [
              8.6e-05,
              4.3e-05,
              2.9e-05,
              1.4e-05,
              1.9e-05,
              6.4e-05,
              7.3e-05,
              1.4e-05,
              5e-05,
              6.7e-05,
              7.1e-05,
              5.6e-05,
              7.3e-05,
              7.4e-05,
              2.2e-05,
              1e-05,
              8.3e-05,
              9.1e-05,
              8.3e-05,
              6.4e-05,
              9.1e-05,
              5.3e-05,
              1.5e-05,
              2.4e-05,
              5.7e-05,
              3.7e-05,
              3.6e-05,
              5e-05,
              1.9e-05,
              6.3e-05,
              4.7e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005546708001929801,
            "speedup_vs_optimized_serial": 0.058067375886524816
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.2417e-05,
            "mad_seconds": 7.417e-06,
            "median_process_cpu_seconds": 0.00014,
            "median_process_cpu_wall_ratio": 1.907692307692308,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 13808.912271980336,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              6.5e-05,
              6.0583e-05,
              7.6709e-05,
              8.0417e-05,
              6.2084e-05,
              6.9042e-05,
              7.7792e-05,
              7.0084e-05,
              6.3791e-05,
              7.2417e-05,
              7.3167e-05,
              6.8459e-05,
              6.3792e-05,
              6.6792e-05,
              6.375e-05,
              7.0042e-05,
              9.2542e-05,
              7.2791e-05,
              5.6333e-05,
              7.0541e-05,
              7.3458e-05,
              7.1583e-05,
              8.975e-05,
              9.0833e-05,
              8.6291e-05,
              6.0542e-05,
              8.0833e-05,
              7.6542e-05,
              8.1209e-05,
              7.2667e-05,
              0.000104625
            ],
            "raw_process_cpu_seconds": [
              0.000124,
              0.000126,
              0.000135,
              0.000166,
              0.000129,
              0.000116,
              0.000152,
              0.000121,
              0.000121,
              0.000132,
              0.000159,
              0.000132,
              0.000124,
              0.000136,
              0.000111,
              0.000145,
              0.000169,
              0.000143,
              9.4e-05,
              0.000143,
              0.00014,
              0.00013,
              0.000159,
              0.000161,
              0.000157,
              0.000113,
              0.000168,
              0.00014,
              0.00017,
              0.000141,
              0.000222
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.007364042001427151,
            "speedup_vs_optimized_serial": 0.012662772553405968
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "final-wide",
      "kernel": "final-borrowed-stop-token",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:21:54Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.179635083,
            "mad_seconds": 0.0009268749999999937,
            "median_process_cpu_seconds": 0.179325,
            "median_process_cpu_wall_ratio": 0.9983861148814114,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 356.27784356578053,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.179188875,
              0.179641208,
              0.180623625,
              0.18074,
              0.179635083,
              0.178708208,
              0.178425959
            ],
            "raw_process_cpu_seconds": [
              0.178882,
              0.179381,
              0.180338,
              0.180428,
              0.179325,
              0.178514,
              0.178138
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.340398207998078
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.178957042,
            "mad_seconds": 0.0005442079999999905,
            "median_process_cpu_seconds": 0.178684,
            "median_process_cpu_wall_ratio": 0.9985249728525416,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 357.6277260997642,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.178957042,
              0.178485375,
              0.177781958,
              0.1784735,
              0.179940417,
              0.17950125,
              0.180324
            ],
            "raw_process_cpu_seconds": [
              0.178684,
              0.178279,
              0.177562,
              0.178098,
              0.179675,
              0.179269,
              0.179979
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.337123999997857,
            "speedup_vs_original_serial": 1.0037888478286314
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.176491958,
            "mad_seconds": 0.0010579169999999916,
            "median_process_cpu_seconds": 0.176258,
            "median_process_cpu_wall_ratio": 0.9984870894804911,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 362.62275474330676,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.1785425,
              0.178007708,
              0.176491958,
              0.175462459,
              0.177549875,
              0.175248916,
              0.176481458
            ],
            "raw_process_cpu_seconds": [
              0.17827,
              0.177738,
              0.176258,
              0.175197,
              0.177315,
              0.174922,
              0.176243
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 3.037459665996721,
            "speedup_vs_optimized_serial": 1.0139671179805259,
            "speedup_vs_original_serial": 1.0178088850937899
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.176766792,
            "mad_seconds": 0.00038462499999999955,
            "median_process_cpu_seconds": 0.176486,
            "median_process_cpu_wall_ratio": 0.9987204538917048,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 362.0589550553138,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.177171416,
              0.176382167,
              0.177637875,
              0.1772615,
              0.176766792,
              0.17642975,
              0.176581417
            ],
            "raw_process_cpu_seconds": [
              0.176974,
              0.176166,
              0.177403,
              0.177016,
              0.176486,
              0.176204,
              0.176375
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.3170840829989174,
            "speedup_vs_optimized_serial": 1.0123906191610923,
            "speedup_vs_original_serial": 1.0162264131602274
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.088475292,
            "mad_seconds": 0.00034933299999999334,
            "median_process_cpu_seconds": 0.176501,
            "median_process_cpu_wall_ratio": 1.9949185361264477,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 723.3657957297276,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.089042542,
              0.089031167,
              0.08837175,
              0.088475292,
              0.0882855,
              0.088839834,
              0.088125959
            ],
            "raw_process_cpu_seconds": [
              0.177587,
              0.177509,
              0.176428,
              0.176501,
              0.17601,
              0.177253,
              0.175907
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.972933709002973,
            "speedup_vs_optimized_serial": 2.0226781732463794,
            "speedup_vs_original_serial": 2.0303417930511043
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.088573333,
            "mad_seconds": 6.370800000000953e-05,
            "median_process_cpu_seconds": 0.176617,
            "median_process_cpu_wall_ratio": 1.9930784586313752,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 722.5651088460226,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.087676083,
              0.088587417,
              0.088516333,
              0.089418667,
              0.088573333,
              0.089105875,
              0.088509625
            ],
            "raw_process_cpu_seconds": [
              0.174985,
              0.176665,
              0.176191,
              0.177482,
              0.176517,
              0.177595,
              0.176617
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.6132175000020652,
            "speedup_vs_optimized_serial": 2.020439289554566,
            "speedup_vs_original_serial": 2.0280944265696763
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.044953209,
            "mad_seconds": 0.0003648340000000014,
            "median_process_cpu_seconds": 0.177917,
            "median_process_cpu_wall_ratio": 3.9578264590632446,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1423.7025881734048,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.045103084,
              0.047251625,
              0.044588375,
              0.044662875,
              0.044953209,
              0.046918875,
              0.044419416
            ],
            "raw_process_cpu_seconds": [
              0.178251,
              0.18087,
              0.176955,
              0.177242,
              0.177917,
              0.180086,
              0.177214
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.4506900829946971,
            "speedup_vs_optimized_serial": 3.9809625604258865,
            "speedup_vs_original_serial": 3.996045821778819
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.044880167,
            "mad_seconds": 0.00015358399999999828,
            "median_process_cpu_seconds": 0.177719,
            "median_process_cpu_wall_ratio": 3.9598560317300064,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1426.019649169309,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.044473292,
              0.044880167,
              0.044752209,
              0.044726583,
              0.045140625,
              0.047288708,
              0.044949958
            ],
            "raw_process_cpu_seconds": [
              0.177167,
              0.177719,
              0.177234,
              0.177372,
              0.177831,
              0.180519,
              0.177745
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.2691278749989579,
            "speedup_vs_optimized_serial": 3.9874415351440207,
            "speedup_vs_original_serial": 4.002549344346246
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.180978875,
            "mad_seconds": 0.0011446250000000102,
            "median_process_cpu_seconds": 0.180463,
            "median_process_cpu_wall_ratio": 0.997932065898815,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 353.63243361966965,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.1870025,
              0.181068417,
              0.17983425,
              0.178899583,
              0.179950125,
              0.180978875,
              0.188914792
            ],
            "raw_process_cpu_seconds": [
              0.18446,
              0.180728,
              0.17961,
              0.17863,
              0.179578,
              0.180463,
              0.188292
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.371693499997491
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.179085083,
            "mad_seconds": 0.0012685420000000114,
            "median_process_cpu_seconds": 0.178658,
            "median_process_cpu_wall_ratio": 0.9977633998030659,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 357.3720319296499,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.177129625,
              0.177816541,
              0.178489209,
              0.180029958,
              0.198437584,
              0.1849935,
              0.179085083
            ],
            "raw_process_cpu_seconds": [
              0.176913,
              0.177521,
              0.17809,
              0.179552,
              0.192419,
              0.184645,
              0.178658
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 3.1451677500008373,
            "speedup_vs_optimized_serial": 1.0105748171108144
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.178658916,
            "mad_seconds": 0.0006666670000000097,
            "median_process_cpu_seconds": 0.178394,
            "median_process_cpu_wall_ratio": 0.9971913457026189,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 358.22449521634843,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.182295125,
              0.181775208,
              0.179325583,
              0.17846625,
              0.178658916,
              0.1772935,
              0.178584625
            ],
            "raw_process_cpu_seconds": [
              0.180889,
              0.181078,
              0.178579,
              0.177965,
              0.178268,
              0.177018,
              0.178394
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.3475519999992684,
            "speedup_vs_optimized_serial": 1.0129854084640255
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.089460625,
            "mad_seconds": 0.00037783400000000744,
            "median_process_cpu_seconds": 0.178158,
            "median_process_cpu_wall_ratio": 1.9927754620979492,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 715.3985342713623,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.089609416,
              0.091706875,
              0.088966958,
              0.089082791,
              0.089460625,
              0.088995125,
              0.089827334
            ],
            "raw_process_cpu_seconds": [
              0.178158,
              0.1792,
              0.177577,
              0.177522,
              0.178505,
              0.17754,
              0.178373
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.0024298339994857,
            "speedup_vs_optimized_serial": 2.023000342329377
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.090130458,
            "mad_seconds": 0.0001412079999999899,
            "median_process_cpu_seconds": 0.17859,
            "median_process_cpu_wall_ratio": 1.9787263037285898,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 710.0818238380638,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.08998925,
              0.09024775,
              0.090130458,
              0.091843,
              0.0897405,
              0.094037791,
              0.090119083
            ],
            "raw_process_cpu_seconds": [
              0.17859,
              0.178194,
              0.178601,
              0.179216,
              0.17801,
              0.179665,
              0.178321
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.6397981670015724,
            "speedup_vs_optimized_serial": 2.007965775565015
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.05156625,
            "mad_seconds": 0.002906416000000002,
            "median_process_cpu_seconds": 0.190486,
            "median_process_cpu_wall_ratio": 3.575142639444728,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1241.1218578042808,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.055858,
              0.057397542,
              0.05156625,
              0.056590583,
              0.048659834,
              0.049778708,
              0.051288583
            ],
            "raw_process_cpu_seconds": [
              0.195555,
              0.191962,
              0.190486,
              0.191696,
              0.182937,
              0.184578,
              0.183364
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.5502597090016934,
            "speedup_vs_optimized_serial": 3.5096380869270116
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.045182666,
            "mad_seconds": 0.0005420000000000008,
            "median_process_cpu_seconds": 0.177794,
            "median_process_cpu_wall_ratio": 3.929836462444699,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1416.4724144431848,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.045182666,
              0.045910541,
              0.0457885,
              0.044640666,
              0.044865416,
              0.045097042,
              0.046235708
            ],
            "raw_process_cpu_seconds": [
              0.177794,
              0.178178,
              0.179266,
              0.177284,
              0.177687,
              0.177224,
              0.178335
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.3089915829987149,
            "speedup_vs_optimized_serial": 4.005493500538459
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "final-batch-k32",
      "kernel": "final-borrowed-stop-token",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:24:44Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.055652375,
            "mad_seconds": 0.0001527500000000001,
            "median_process_cpu_seconds": 0.055584,
            "median_process_cpu_wall_ratio": 0.9990142052982443,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4599.983378966307,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.055985875,
              0.055750959,
              0.056203083,
              0.055648083,
              0.055652375,
              0.055499625,
              0.055430625
            ],
            "raw_process_cpu_seconds": [
              0.055955,
              0.055696,
              0.05616,
              0.055568,
              0.055584,
              0.05543,
              0.055379
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7555589580006199
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057567833,
            "mad_seconds": 0.0002912909999999991,
            "median_process_cpu_seconds": 0.057485,
            "median_process_cpu_wall_ratio": 0.9989796908873237,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4446.927852920919,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056938125,
              0.057567833,
              0.057276542,
              0.057948125,
              0.057894625,
              0.057636291,
              0.057414375
            ],
            "raw_process_cpu_seconds": [
              0.056889,
              0.057485,
              0.057131,
              0.057889,
              0.05786,
              0.057589,
              0.057324
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7985224999938509,
            "speedup_vs_original_serial": 0.9667269393308586
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057382334,
            "mad_seconds": 0.0003276659999999973,
            "median_process_cpu_seconds": 0.057293,
            "median_process_cpu_wall_ratio": 0.9990534910258565,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4461.3033690821985,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.0565235,
              0.0571355,
              0.05771,
              0.057401833,
              0.057382334,
              0.058539292,
              0.056503083
            ],
            "raw_process_cpu_seconds": [
              0.05647,
              0.057093,
              0.057647,
              0.057293,
              0.057343,
              0.058474,
              0.056463
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.002316999998584,
            "speedup_vs_optimized_serial": 1.0032326848189896,
            "speedup_vs_original_serial": 0.9698520628317419
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.056621208,
            "mad_seconds": 0.00013279100000000016,
            "median_process_cpu_seconds": 0.056587,
            "median_process_cpu_wall_ratio": 0.9988419395785157,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4521.274078080425,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056621208,
              0.056695542,
              0.056488417,
              0.056809375,
              0.056311667,
              0.05660025,
              0.056792833
            ],
            "raw_process_cpu_seconds": [
              0.056587,
              0.056648,
              0.056423,
              0.05674,
              0.056232,
              0.05654,
              0.05672
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7633712919996469,
            "speedup_vs_optimized_serial": 1.0167185588834486,
            "speedup_vs_original_serial": 0.9828892205902777
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028526333,
            "mad_seconds": 7.766600000000054e-05,
            "median_process_cpu_seconds": 0.056745,
            "median_process_cpu_wall_ratio": 1.991154985687032,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8974.164327395323,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.028206375,
              0.028448667,
              0.029856792,
              0.028464459,
              0.028584917,
              0.02875775,
              0.028526333
            ],
            "raw_process_cpu_seconds": [
              0.056289,
              0.056664,
              0.056906,
              0.056745,
              0.056917,
              0.057144,
              0.056585
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.6540590830045403,
            "speedup_vs_optimized_serial": 2.0180593488830127,
            "speedup_vs_original_serial": 1.9509123377337
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028691791,
            "mad_seconds": 0.00012987599999999752,
            "median_process_cpu_seconds": 0.057178,
            "median_process_cpu_wall_ratio": 1.992834814668767,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8922.41268591424,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.028536917,
              0.029038792,
              0.028666709,
              0.028821667,
              0.028691791,
              0.028560458,
              0.028704542
            ],
            "raw_process_cpu_seconds": [
              0.0569,
              0.057736,
              0.057182,
              0.057322,
              0.057178,
              0.057036,
              0.056977
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.5415078750011162,
            "speedup_vs_optimized_serial": 2.0064217322648137,
            "speedup_vs_original_serial": 1.939661940239283
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014485583,
            "mad_seconds": 3.099999999999978e-05,
            "median_process_cpu_seconds": 0.057427,
            "median_process_cpu_wall_ratio": 3.963736160244775,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17672.743996565412,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014454583,
              0.01461275,
              0.015103917,
              0.014485583,
              0.014465375,
              0.014472709,
              0.014599583
            ],
            "raw_process_cpu_seconds": [
              0.057427,
              0.057662,
              0.058628,
              0.057163,
              0.057354,
              0.057366,
              0.058002
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.49100449999968987,
            "speedup_vs_optimized_serial": 3.9741467775235555,
            "speedup_vs_original_serial": 3.8419147506869415
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.0144945,
            "mad_seconds": 2.7583000000001162e-05,
            "median_process_cpu_seconds": 0.057293,
            "median_process_cpu_wall_ratio": 3.948235988509377,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17661.87174445479,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014422875,
              0.014600541,
              0.0144945,
              0.014503792,
              0.014466917,
              0.014482417,
              0.014537792
            ],
            "raw_process_cpu_seconds": [
              0.057251,
              0.057448,
              0.057336,
              0.057193,
              0.057293,
              0.05718,
              0.057384
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.4305320839994238,
            "speedup_vs_optimized_serial": 3.9717018869226255,
            "speedup_vs_original_serial": 3.8395512090793056
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057452875,
            "mad_seconds": 0.00018887499999999807,
            "median_process_cpu_seconds": 0.057338,
            "median_process_cpu_wall_ratio": 0.9990680006393505,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4455.825752845267,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057264,
              0.057859167,
              0.057314958,
              0.0581835,
              0.057314417,
              0.057827625,
              0.057452875
            ],
            "raw_process_cpu_seconds": [
              0.057225,
              0.057783,
              0.05727,
              0.058134,
              0.057261,
              0.057699,
              0.057338
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7717994579943479
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.057302583,
            "mad_seconds": 0.0004100409999999999,
            "median_process_cpu_seconds": 0.057243,
            "median_process_cpu_wall_ratio": 0.9982099774301502,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4467.512398175838,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056892542,
              0.05705775,
              0.0578205,
              0.057467291,
              0.058240042,
              0.056223875,
              0.057302583
            ],
            "raw_process_cpu_seconds": [
              0.056766,
              0.056976,
              0.057717,
              0.05735,
              0.057945,
              0.056164,
              0.057243
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0045191250028438,
            "speedup_vs_optimized_serial": 1.0026227788021354
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.056781708,
            "mad_seconds": 4.2624999999997526e-05,
            "median_process_cpu_seconds": 0.056718,
            "median_process_cpu_wall_ratio": 0.9993816784102767,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4508.494179146566,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056807541,
              0.056752458,
              0.056418167,
              0.057218916,
              0.057095042,
              0.056739083,
              0.056781708
            ],
            "raw_process_cpu_seconds": [
              0.056718,
              0.056658,
              0.056254,
              0.057202,
              0.057077,
              0.056704,
              0.056755
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7684916250000242,
            "speedup_vs_optimized_serial": 1.0118201270028722
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028508292,
            "mad_seconds": 0.00011791599999999916,
            "median_process_cpu_seconds": 0.056723,
            "median_process_cpu_wall_ratio": 1.9931663464297205,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8979.843478521969,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.028339,
              0.028420209,
              0.028508292,
              0.028543584,
              0.028626208,
              0.02838775,
              0.02875475
            ],
            "raw_process_cpu_seconds": [
              0.056536,
              0.056723,
              0.056623,
              0.05699,
              0.057026,
              0.056541,
              0.057313
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.6554333749954822,
            "speedup_vs_optimized_serial": 2.0153040034808116
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028874125,
            "mad_seconds": 0.0001991670000000001,
            "median_process_cpu_seconds": 0.057503,
            "median_process_cpu_wall_ratio": 1.9873305313507539,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8866.06953457464,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.028674958,
              0.028874125,
              0.02892725,
              0.030195417,
              0.028539792,
              0.028952833,
              0.028243125
            ],
            "raw_process_cpu_seconds": [
              0.056971,
              0.057646,
              0.057696,
              0.057573,
              0.056718,
              0.057503,
              0.056375
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.543810167000629,
            "speedup_vs_optimized_serial": 1.9897702527782226
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014574125,
            "mad_seconds": 1.9207999999999725e-05,
            "median_process_cpu_seconds": 0.057845,
            "median_process_cpu_wall_ratio": 3.959140929458016,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17565.37699518839,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014690042,
              0.014593333,
              0.014574125,
              0.014624833,
              0.01456225,
              0.01449575,
              0.014565666
            ],
            "raw_process_cpu_seconds": [
              0.057959,
              0.058093,
              0.057845,
              0.057863,
              0.057654,
              0.057606,
              0.057653
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.4845322499968461,
            "speedup_vs_optimized_serial": 3.942114878251696
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014644,
            "mad_seconds": 6.904200000000006e-05,
            "median_process_cpu_seconds": 0.057975,
            "median_process_cpu_wall_ratio": 3.9565288026609435,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17481.562414640808,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014738834,
              0.014618875,
              0.014547333,
              0.014644,
              0.014585917,
              0.014713042,
              0.0148135
            ],
            "raw_process_cpu_seconds": [
              0.057948,
              0.05784,
              0.057225,
              0.05811,
              0.057975,
              0.058348,
              0.058348
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.4337679579984979,
            "speedup_vs_optimized_serial": 3.923304766457252
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "final-all-neighbors",
      "kernel": "final-borrowed-stop-token",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:24:55Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "8336a6fcde6abf56b9772887771896e1375208d35dd13e5041ac00b3241d6ea8"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.001996167,
            "mad_seconds": 3.095900000000011e-05,
            "median_process_cpu_seconds": 0.001997,
            "median_process_cpu_wall_ratio": 0.9998533615361284,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 128245.78304320229,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.001956417,
              0.002009875,
              0.002091292,
              0.002046709,
              0.001991292,
              0.00194775,
              0.001958959,
              0.002131333,
              0.001996167,
              0.002008584,
              0.002024875,
              0.001984375,
              0.001973875,
              0.002036916,
              0.001965208
            ],
            "raw_process_cpu_seconds": [
              0.001959,
              0.002007,
              0.002084,
              0.00204,
              0.001991,
              0.001948,
              0.001952,
              0.002132,
              0.001997,
              0.002008,
              0.002014,
              0.001985,
              0.001973,
              0.002037,
              0.001965
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.04821845900005428
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000197541,
            "mad_seconds": 4.12399999999999e-06,
            "median_process_cpu_seconds": 0.000199,
            "median_process_cpu_wall_ratio": 1.0018785222291797,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1295933.5024121576,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000196083,
              0.000190875,
              0.000193417,
              0.000212041,
              0.000202042,
              0.000196792,
              0.000198708,
              0.000196792,
              0.000194292,
              0.000190584,
              0.000197541,
              0.000215166,
              0.000224375,
              0.000199625,
              0.000218
            ],
            "raw_process_cpu_seconds": [
              0.000199,
              0.000191,
              0.000193,
              0.000212,
              0.000203,
              0.000197,
              0.0002,
              0.000197,
              0.000196,
              0.000191,
              0.000198,
              0.000215,
              0.000225,
              0.0002,
              0.000218
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.015251042001182213,
            "speedup_vs_original_serial": 10.105076920740505
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.00022225,
            "mad_seconds": 2.7919999999999963e-06,
            "median_process_cpu_seconds": 0.000227,
            "median_process_cpu_wall_ratio": 1.0225456422678671,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1151856.0179977503,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.00023325,
              0.0002265,
              0.000211458,
              0.000223,
              0.000224459,
              0.000223041,
              0.00022225,
              0.000218709,
              0.000201458,
              0.000200166,
              0.000221667,
              0.00022175,
              0.000237542,
              0.000224459,
              0.000219458
            ],
            "raw_process_cpu_seconds": [
              0.00024,
              0.000238,
              0.000217,
              0.000233,
              0.000229,
              0.000229,
              0.000227,
              0.000223,
              0.000206,
              0.000204,
              0.000227,
              0.000226,
              0.000242,
              0.000228,
              0.000226
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.021206917001109105,
            "speedup_vs_optimized_serial": 0.8888233970753655,
            "speedup_vs_original_serial": 8.98162879640045
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000244667,
            "mad_seconds": 6.000999999999999e-06,
            "median_process_cpu_seconds": 0.000248,
            "median_process_cpu_wall_ratio": 1.0214592274678111,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1046320.1003813345,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000256792,
              0.000244958,
              0.000240625,
              0.000244667,
              0.000233,
              0.000239291,
              0.000240042,
              0.000240958,
              0.000238666,
              0.000257917,
              0.000255417,
              0.000261125,
              0.000232375,
              0.000244958,
              0.000257125
            ],
            "raw_process_cpu_seconds": [
              0.000268,
              0.000251,
              0.000247,
              0.000253,
              0.000238,
              0.000243,
              0.000245,
              0.000245,
              0.000243,
              0.000262,
              0.000259,
              0.000273,
              0.000238,
              0.000248,
              0.000263
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.016526332998182625,
            "speedup_vs_optimized_serial": 0.8073871833962079,
            "speedup_vs_original_serial": 8.1587095930387
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000113375,
            "mad_seconds": 4.2500000000000084e-06,
            "median_process_cpu_seconds": 0.000227,
            "median_process_cpu_wall_ratio": 1.9976553341148886,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 2257993.3847850054,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000126625,
              0.000113375,
              0.000108875,
              0.000113083,
              0.000115292,
              0.000113417,
              0.000109125,
              0.0001195,
              0.000117458,
              0.000119209,
              0.000112291,
              0.000106625,
              0.00011475,
              0.000105208,
              0.000106125
            ],
            "raw_process_cpu_seconds": [
              0.000244,
              0.000227,
              0.00022,
              0.000229,
              0.000226,
              0.000228,
              0.000216,
              0.000238,
              0.000228,
              0.000231,
              0.000224,
              0.000213,
              0.000231,
              0.000212,
              0.000214
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.016903000003367197,
            "speedup_vs_optimized_serial": 1.7423682469680264,
            "speedup_vs_original_serial": 17.606765159867695
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000152125,
            "mad_seconds": 6.709000000000006e-06,
            "median_process_cpu_seconds": 0.000276,
            "median_process_cpu_wall_ratio": 1.8142974527526705,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1682826.6228430567,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000168666,
              0.000158834,
              0.000175708,
              0.000152125,
              0.000147792,
              0.000186625,
              0.000148916,
              0.000161042,
              0.000155792,
              0.00014525,
              0.000143667,
              0.000149792,
              0.000145667,
              0.000147334,
              0.000163917
            ],
            "raw_process_cpu_seconds": [
              0.0003,
              0.000292,
              0.000302,
              0.000276,
              0.000275,
              0.000329,
              0.000274,
              0.000288,
              0.000283,
              0.000262,
              0.000267,
              0.000274,
              0.00026,
              0.000269,
              0.000297
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.014724166998348664,
            "speedup_vs_optimized_serial": 1.298543960558751,
            "speedup_vs_original_serial": 13.121886606409204
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 8.725e-05,
            "mad_seconds": 5.583000000000004e-06,
            "median_process_cpu_seconds": 0.00031,
            "median_process_cpu_wall_ratio": 3.5603272549507916,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 2934097.421203438,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              8.4875e-05,
              8.1667e-05,
              0.000106958,
              7.8542e-05,
              8.6959e-05,
              0.000117333,
              8.8291e-05,
              9.1917e-05,
              8.725e-05,
              8.5125e-05,
              9.85e-05,
              9.1958e-05,
              6.9625e-05,
              0.000108417,
              7.0417e-05
            ],
            "raw_process_cpu_seconds": [
              0.000286,
              0.000275,
              0.000388,
              0.000289,
              0.000321,
              0.000402,
              0.000337,
              0.000308,
              0.000332,
              0.000283,
              0.000341,
              0.00031,
              0.00025,
              0.000386,
              0.000257
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.015337917000579182,
            "speedup_vs_optimized_serial": 2.264080229226361,
            "speedup_vs_original_serial": 22.878704871060172
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000167209,
            "mad_seconds": 8.792000000000007e-06,
            "median_process_cpu_seconds": 0.000439,
            "median_process_cpu_wall_ratio": 2.644578313253012,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1531018.0672093008,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.00020225,
              0.000169291,
              0.000188417,
              0.000172125,
              0.000167209,
              0.000144667,
              0.000156709,
              0.000166,
              0.000172875,
              0.000158417,
              0.000168667,
              0.000165083,
              0.000141917,
              0.00018875,
              0.000148958
            ],
            "raw_process_cpu_seconds": [
              0.000507,
              0.000478,
              0.000505,
              0.000451,
              0.000424,
              0.000383,
              0.000387,
              0.000439,
              0.000472,
              0.000417,
              0.000451,
              0.000432,
              0.000384,
              0.000461,
              0.00042
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.015654999995604157,
            "speedup_vs_optimized_serial": 1.181401718807002,
            "speedup_vs_original_serial": 11.9381552428398
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.002026667,
            "mad_seconds": 2.1124999999999877e-05,
            "median_process_cpu_seconds": 0.002027,
            "median_process_cpu_wall_ratio": 0.9998355825930847,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 126315.76869806438,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.002003083,
              0.002028833,
              0.002020375,
              0.002047667,
              0.002005542,
              0.002026667,
              0.001997833,
              0.002014792,
              0.002026958,
              0.001995958,
              0.002136458,
              0.002054,
              0.002129041,
              0.002025333,
              0.002100959
            ],
            "raw_process_cpu_seconds": [
              0.002005,
              0.00203,
              0.00202,
              0.002047,
              0.002005,
              0.002027,
              0.001998,
              0.002015,
              0.002027,
              0.001996,
              0.002134,
              0.002045,
              0.002126,
              0.002025,
              0.0021
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.04943033400195418
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.002071334,
            "mad_seconds": 3.0251000000000236e-05,
            "median_process_cpu_seconds": 0.002073,
            "median_process_cpu_wall_ratio": 1.0025574543485316,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 123591.84950374975,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.002162208,
              0.002069958,
              0.002041083,
              0.00204,
              0.002185625,
              0.002083375,
              0.002006041,
              0.00220475,
              0.002071334,
              0.002060458,
              0.002090209,
              0.002036791,
              0.002058459,
              0.002112458,
              0.00207575
            ],
            "raw_process_cpu_seconds": [
              0.002172,
              0.002074,
              0.002047,
              0.002047,
              0.002185,
              0.002093,
              0.002009,
              0.002198,
              0.002072,
              0.002065,
              0.002097,
              0.002042,
              0.002064,
              0.002119,
              0.002073
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.0589578330036602,
            "speedup_vs_optimized_serial": 0.9784356361649061
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.002112583,
            "mad_seconds": 2.2334000000000156e-05,
            "median_process_cpu_seconds": 0.002116,
            "median_process_cpu_wall_ratio": 1.0004410596111861,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 121178.67084985538,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.002076834,
              0.002110875,
              0.002172042,
              0.002131875,
              0.002078333,
              0.002104417,
              0.002135625,
              0.002179,
              0.002112583,
              0.00210825,
              0.002095167,
              0.002092125,
              0.002134917,
              0.002204333,
              0.002145833
            ],
            "raw_process_cpu_seconds": [
              0.002087,
              0.002116,
              0.002173,
              0.002131,
              0.002083,
              0.00211,
              0.002139,
              0.002177,
              0.002113,
              0.002112,
              0.002106,
              0.002091,
              0.002132,
              0.002205,
              0.002143
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.050422874999640044,
            "speedup_vs_optimized_serial": 0.9593313020127493
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.001090208,
            "mad_seconds": 1.2249999999999891e-05,
            "median_process_cpu_seconds": 0.00215,
            "median_process_cpu_wall_ratio": 1.9698288184701471,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 234817.57609557075,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.001102458,
              0.001105292,
              0.001079208,
              0.001115833,
              0.00105525,
              0.001130167,
              0.001078542,
              0.001147084,
              0.0010845,
              0.001064416,
              0.00108975,
              0.001105166,
              0.0010915,
              0.001090208,
              0.001087375
            ],
            "raw_process_cpu_seconds": [
              0.002134,
              0.002174,
              0.002145,
              0.002198,
              0.002089,
              0.00222,
              0.002139,
              0.0022,
              0.002032,
              0.002121,
              0.00215,
              0.002162,
              0.002133,
              0.002152,
              0.002151
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.038802000002760906,
            "speedup_vs_optimized_serial": 1.8589727831753204
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.001126291,
            "mad_seconds": 2.358299999999998e-05,
            "median_process_cpu_seconds": 0.002203,
            "median_process_cpu_wall_ratio": 1.9508395498497353,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 227294.72223430712,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.001204542,
              0.001136,
              0.001167375,
              0.001089125,
              0.001202666,
              0.00112425,
              0.001116167,
              0.001102708,
              0.001132333,
              0.00113,
              0.001055416,
              0.001090583,
              0.001186,
              0.001126291,
              0.001116542
            ],
            "raw_process_cpu_seconds": [
              0.002284,
              0.002223,
              0.002245,
              0.002148,
              0.002339,
              0.002222,
              0.002203,
              0.002163,
              0.002209,
              0.002154,
              0.002089,
              0.002154,
              0.00225,
              0.00218,
              0.002171
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.035323042000527494,
            "speedup_vs_optimized_serial": 1.7994168469782674
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000570917,
            "mad_seconds": 1.6416e-05,
            "median_process_cpu_seconds": 0.002212,
            "median_process_cpu_wall_ratio": 3.8744686180302916,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 448401.4313814442,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000587333,
              0.000597,
              0.000571708,
              0.00060425,
              0.000584708,
              0.000570917,
              0.000602375,
              0.000551209,
              0.000552541,
              0.000584458,
              0.000570792,
              0.000546083,
              0.000546417,
              0.000555042,
              0.000569709
            ],
            "raw_process_cpu_seconds": [
              0.002278,
              0.002268,
              0.002212,
              0.002354,
              0.002232,
              0.002212,
              0.002227,
              0.002158,
              0.002146,
              0.002231,
              0.002189,
              0.002133,
              0.002126,
              0.002167,
              0.0022
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.029050791999907233,
            "speedup_vs_optimized_serial": 3.54984524895913
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000650708,
            "mad_seconds": 2.3957999999999922e-05,
            "median_process_cpu_seconds": 0.002343,
            "median_process_cpu_wall_ratio": 3.648982848025528,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 393417.6312570308,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000656208,
              0.00062675,
              0.000624667,
              0.000601458,
              0.000617416,
              0.000650708,
              0.000614166,
              0.000616792,
              0.000638333,
              0.000680958,
              0.00065225,
              0.000709208,
              0.00067075,
              0.000652416,
              0.000664167
            ],
            "raw_process_cpu_seconds": [
              0.002381,
              0.002287,
              0.002281,
              0.00224,
              0.002306,
              0.002359,
              0.002268,
              0.002275,
              0.002336,
              0.002415,
              0.002354,
              0.002445,
              0.002452,
              0.002371,
              0.002343
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.027291749996948056,
            "speedup_vs_optimized_serial": 3.1145567597140347
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "integrated-batch",
      "kernel": "final-borrowed-stop-token-with-concurrent-source-integration",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:35:03Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "71cdb79f8d322e636bf6a595bbd4bf92f5593a66e62ade3ae7c12cbafd776052"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.055967125,
            "mad_seconds": 0.00033249999999999946,
            "median_process_cpu_seconds": 0.055937,
            "median_process_cpu_wall_ratio": 0.9992585081332148,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4574.113821283477,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.055249708,
              0.056299625,
              0.056023541,
              0.055967125,
              0.054161916,
              0.055383166,
              0.055304459,
              0.056287041,
              0.056026875
            ],
            "raw_process_cpu_seconds": [
              0.055169,
              0.05627,
              0.055982,
              0.055937,
              0.054144,
              0.055284,
              0.055274,
              0.056237,
              0.055943
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0830905830007396
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.05537575,
            "mad_seconds": 0.0004780839999999967,
            "median_process_cpu_seconds": 0.055352,
            "median_process_cpu_wall_ratio": 0.9991895770761515,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4622.96221721602,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.055488459,
              0.056265125,
              0.054692708,
              0.054259708,
              0.05537575,
              0.055853834,
              0.055115667,
              0.056036083,
              0.055269125
            ],
            "raw_process_cpu_seconds": [
              0.055435,
              0.055923,
              0.054628,
              0.054239,
              0.055352,
              0.055817,
              0.055071,
              0.056014,
              0.055182
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.048679292005545,
            "speedup_vs_original_serial": 1.0106793135984613
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.054808792,
            "mad_seconds": 0.00048249899999999735,
            "median_process_cpu_seconds": 0.05478,
            "median_process_cpu_wall_ratio": 0.9994746828209605,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4670.783475760604,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.054982375,
              0.053446583,
              0.054248459,
              0.053769709,
              0.054808792,
              0.054947375,
              0.0548495,
              0.0538655,
              0.055291291
            ],
            "raw_process_cpu_seconds": [
              0.054955,
              0.053391,
              0.054234,
              0.053723,
              0.05478,
              0.054919,
              0.054823,
              0.053826,
              0.055223
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0702832499955548,
            "speedup_vs_optimized_serial": 1.0103442892884777,
            "speedup_vs_original_serial": 1.021134072796204
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.0549205,
            "mad_seconds": 0.00046116699999999844,
            "median_process_cpu_seconds": 0.05493,
            "median_process_cpu_wall_ratio": 0.9997867109407345,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4661.283127429649,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056049125,
              0.054875125,
              0.054459333,
              0.0549205,
              0.05455125,
              0.054306583,
              0.055143,
              0.056092917,
              0.05566575
            ],
            "raw_process_cpu_seconds": [
              0.056058,
              0.054852,
              0.054442,
              0.05493,
              0.05446,
              0.054295,
              0.055116,
              0.056092,
              0.055672
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8576769169958425,
            "speedup_vs_optimized_serial": 1.0082892544678217,
            "speedup_vs_original_serial": 1.0190570916142425
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.027862958,
            "mad_seconds": 0.0004040000000000016,
            "median_process_cpu_seconds": 0.055581,
            "median_process_cpu_wall_ratio": 1.9788806261330092,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 9187.825642919894,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.027800834,
              0.028302417,
              0.027458958,
              0.027288166,
              0.028483834,
              0.027919375,
              0.027862958,
              0.027845666,
              0.028697125
            ],
            "raw_process_cpu_seconds": [
              0.054911,
              0.055898,
              0.054338,
              0.054492,
              0.056353,
              0.05579,
              0.055454,
              0.055581,
              0.056488
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.702216124998813,
            "speedup_vs_optimized_serial": 1.9874325618981301,
            "speedup_vs_original_serial": 2.008656977482434
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028463834,
            "mad_seconds": 0.00032616599999999926,
            "median_process_cpu_seconds": 0.056531,
            "median_process_cpu_wall_ratio": 1.986134404802951,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8993.869202581775,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.028463834,
              0.027086917,
              0.02879,
              0.028229708,
              0.028557541,
              0.028810625,
              0.02885525,
              0.028408542,
              0.027666
            ],
            "raw_process_cpu_seconds": [
              0.056533,
              0.053683,
              0.056468,
              0.056086,
              0.056768,
              0.05662,
              0.056767,
              0.056531,
              0.055004
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.590019791998202,
            "speedup_vs_optimized_serial": 1.9454775488080769,
            "speedup_vs_original_serial": 1.9662539136505643
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014359334,
            "mad_seconds": 4.891599999999954e-05,
            "median_process_cpu_seconds": 0.056738,
            "median_process_cpu_wall_ratio": 3.958361952855759,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17828.124897714617,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014405959,
              0.01440825,
              0.014489208,
              0.014320125,
              0.014172375,
              0.014359334,
              0.014361,
              0.01427875,
              0.01410975
            ],
            "raw_process_cpu_seconds": [
              0.057024,
              0.056849,
              0.057193,
              0.056738,
              0.056378,
              0.056984,
              0.056585,
              0.056475,
              0.056119
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.5051963330042781,
            "speedup_vs_optimized_serial": 3.8564288566586726,
            "speedup_vs_original_serial": 3.897612869789086
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014397625,
            "mad_seconds": 5.7416000000001105e-05,
            "median_process_cpu_seconds": 0.056937,
            "median_process_cpu_wall_ratio": 3.9570552981480254,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17780.71036021566,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014433541,
              0.0142745,
              0.014286792,
              0.01448175,
              0.014416167,
              0.014266042,
              0.014419917,
              0.014340209,
              0.014397625
            ],
            "raw_process_cpu_seconds": [
              0.057036,
              0.056434,
              0.056573,
              0.057233,
              0.057097,
              0.05659,
              0.056937,
              0.056745,
              0.057129
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.45328266699652886,
            "speedup_vs_optimized_serial": 3.8461725458191887,
            "speedup_vs_original_serial": 3.8872470285897847
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.055701625,
            "mad_seconds": 0.0007797920000000014,
            "median_process_cpu_seconds": 0.05567,
            "median_process_cpu_wall_ratio": 0.9994446653740392,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4595.9161873643725,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.056481417,
              0.056513959,
              0.054718,
              0.054547292,
              0.054689458,
              0.055695125,
              0.055906291,
              0.055701625,
              0.056380042
            ],
            "raw_process_cpu_seconds": [
              0.056457,
              0.056486,
              0.054693,
              0.054517,
              0.054653,
              0.05567,
              0.055852,
              0.055625,
              0.056313
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8615478329957114
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.054478125,
            "mad_seconds": 0.0003577500000000039,
            "median_process_cpu_seconds": 0.054404,
            "median_process_cpu_wall_ratio": 0.9987307555069341,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4699.1338266506045,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.054985,
              0.053009833,
              0.054120375,
              0.054536959,
              0.055872209,
              0.055680833,
              0.054478125,
              0.054407583,
              0.054329958
            ],
            "raw_process_cpu_seconds": [
              0.054977,
              0.052958,
              0.054098,
              0.054484,
              0.055793,
              0.055468,
              0.054404,
              0.054304,
              0.054261
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.0705241249961546,
            "speedup_vs_optimized_serial": 1.022458555612918
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.0552845,
            "mad_seconds": 0.0007003750000000031,
            "median_process_cpu_seconds": 0.05528,
            "median_process_cpu_wall_ratio": 0.9986969654638597,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4630.592661595926,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.054584125,
              0.0552845,
              0.055535417,
              0.054983333,
              0.054962791,
              0.054411875,
              0.073623417,
              0.056489542,
              0.05606875
            ],
            "raw_process_cpu_seconds": [
              0.054513,
              0.05528,
              0.05553,
              0.054919,
              0.054965,
              0.054325,
              0.071373,
              0.056408,
              0.055968
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.8797090000007302,
            "speedup_vs_optimized_serial": 1.0075450623592508
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028044666,
            "mad_seconds": 0.00015766600000000075,
            "median_process_cpu_seconds": 0.055683,
            "median_process_cpu_wall_ratio": 1.9803499535633953,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 9128.295555382974,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.02815725,
              0.028652208,
              0.028835334,
              0.027984417,
              0.028044666,
              0.028353333,
              0.027887,
              0.028033459,
              0.027578459
            ],
            "raw_process_cpu_seconds": [
              0.055683,
              0.056644,
              0.056853,
              0.055617,
              0.055553,
              0.05595,
              0.055361,
              0.055932,
              0.054615
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7063457499971264,
            "speedup_vs_optimized_serial": 1.986175374668395
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.027990292,
            "mad_seconds": 0.0002281659999999984,
            "median_process_cpu_seconds": 0.055492,
            "median_process_cpu_wall_ratio": 1.983025711580038,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 9146.028201492147,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.027369791,
              0.028267209,
              0.028721875,
              0.028218458,
              0.027715167,
              0.027819125,
              0.0279835,
              0.027990292,
              0.028198792
            ],
            "raw_process_cpu_seconds": [
              0.054173,
              0.056166,
              0.056803,
              0.056107,
              0.054773,
              0.05525,
              0.055492,
              0.055428,
              0.055926
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.5827906250051456,
            "speedup_vs_optimized_serial": 1.9900337231208591
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014335708,
            "mad_seconds": 0.00010199999999999966,
            "median_process_cpu_seconds": 0.056561,
            "median_process_cpu_wall_ratio": 3.958395360190379,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17857.506584257993,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014278959,
              0.014248334,
              0.014335708,
              0.014165834,
              0.014191125,
              0.014365417,
              0.014466625,
              0.014488166,
              0.014437708
            ],
            "raw_process_cpu_seconds": [
              0.056369,
              0.056456,
              0.056561,
              0.056159,
              0.056156,
              0.056864,
              0.057457,
              0.057386,
              0.056908
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.5039646250006626,
            "speedup_vs_optimized_serial": 3.8855161530912876
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014521292,
            "mad_seconds": 7.158300000000006e-05,
            "median_process_cpu_seconds": 0.057066,
            "median_process_cpu_wall_ratio": 3.942363129942617,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17629.285328054833,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014269625,
              0.014521292,
              0.0145385,
              0.014563125,
              0.014561083,
              0.014401583,
              0.014415667,
              0.014692417,
              0.014449709
            ],
            "raw_process_cpu_seconds": [
              0.056363,
              0.057407,
              0.057138,
              0.057055,
              0.057048,
              0.057066,
              0.057104,
              0.057748,
              0.056966
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "9",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.462849333001941,
            "speedup_vs_optimized_serial": 3.8358587514113758
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "integrated-tiny",
      "kernel": "final-borrowed-stop-token-with-concurrent-source-integration",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:35:15Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "71cdb79f8d322e636bf6a595bbd4bf92f5593a66e62ade3ae7c12cbafd776052"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.08e-07,
            "mad_seconds": 4.100000000000001e-08,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.4124293785310733,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1412429.3785310735,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              7.08e-07,
              7.5e-07,
              7.08e-07,
              7.09e-07,
              6.67e-07,
              1.75e-06,
              7.09e-07,
              7.08e-07,
              6.66e-07,
              7.08e-07,
              7.09e-07,
              7.09e-07,
              7.08e-07,
              7.08e-07,
              7.08e-07,
              7.09e-07,
              6.67e-07,
              7.09e-07,
              6.67e-07,
              6.67e-07,
              6.67e-07,
              6.67e-07,
              6.67e-07,
              7.08e-07,
              6.67e-07,
              6.67e-07,
              6.67e-07,
              6.67e-07,
              6.66e-07,
              7.08e-07,
              6.66e-07
            ],
            "raw_process_cpu_seconds": [
              2e-06,
              0,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              0,
              0,
              0,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005050791005487554
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.92e-07,
            "mad_seconds": 4.100000000000001e-08,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.2626262626262625,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1262626.2626262626,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              8.33e-07,
              1.708e-06,
              8.75e-07,
              8.33e-07,
              1.583e-06,
              8.34e-07,
              7.92e-07,
              8.33e-07,
              7.91e-07,
              7.5e-07,
              7.91e-07,
              7.91e-07,
              7.92e-07,
              7.92e-07,
              7.5e-07,
              7.92e-07,
              7.92e-07,
              7.5e-07,
              7.92e-07,
              7.92e-07,
              7.92e-07,
              7.5e-07,
              7.5e-07,
              7.5e-07,
              7.5e-07,
              7.5e-07,
              7.92e-07,
              7.5e-07,
              7.91e-07,
              7.08e-07,
              7.5e-07
            ],
            "raw_process_cpu_seconds": [
              4e-06,
              2e-06,
              2e-06,
              1e-06,
              2e-06,
              0,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              0,
              0,
              0,
              0,
              0,
              0,
              0,
              0,
              1e-06,
              0,
              0
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005200084000534844,
            "speedup_vs_original_serial": 0.893939393939394
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.167e-06,
            "mad_seconds": 1.7499999999999993e-06,
            "median_process_cpu_seconds": 9e-06,
            "median_process_cpu_wall_ratio": 1.3427324605572337,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 139528.39402818473,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              8.917e-06,
              8.083e-06,
              7.583e-06,
              5.791e-06,
              5.667e-06,
              4.125e-06,
              4.5e-06,
              4.5e-06,
              9.375e-06,
              4.5e-06,
              9.583e-06,
              2.8167e-05,
              7.5e-06,
              7.167e-06,
              6.958e-06,
              7.25e-06,
              5.958e-06,
              6.458e-06,
              5.833e-06,
              4.75e-06,
              4.75e-06,
              5.833e-06,
              8.542e-06,
              9.583e-06,
              8.459e-06,
              1.2666e-05,
              1.1834e-05,
              5.167e-06,
              7.167e-06,
              1.0708e-05,
              5.375e-06
            ],
            "raw_process_cpu_seconds": [
              1.4e-05,
              9e-06,
              1e-05,
              8e-06,
              8e-06,
              6e-06,
              8e-06,
              6e-06,
              1.1e-05,
              7e-06,
              1.2e-05,
              3.3e-05,
              9e-06,
              1e-05,
              8e-06,
              8e-06,
              8e-06,
              9e-06,
              1e-05,
              7e-06,
              7e-06,
              8e-06,
              1.2e-05,
              1.1e-05,
              1e-05,
              1.5e-05,
              1.2e-05,
              7e-06,
              1.1e-05,
              1.4e-05,
              7e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005069541999546345,
            "speedup_vs_optimized_serial": 0.1105064880703223,
            "speedup_vs_original_serial": 0.0987861029719548
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 3.5333e-05,
            "mad_seconds": 1.875000000000005e-06,
            "median_process_cpu_seconds": 3.7e-05,
            "median_process_cpu_wall_ratio": 1.0716602273047746,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 28302.153793903715,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.5125e-05,
              4.0209e-05,
              3.6083e-05,
              3.9917e-05,
              4.3791e-05,
              3.4625e-05,
              3.6459e-05,
              4.0625e-05,
              3.3666e-05,
              3.6334e-05,
              2.8458e-05,
              3.5459e-05,
              3.4167e-05,
              3.5333e-05,
              3.3458e-05,
              3.1875e-05,
              3.8917e-05,
              3.7416e-05,
              3.6417e-05,
              3.3375e-05,
              3.3875e-05,
              4.0208e-05,
              3.3791e-05,
              3.3333e-05,
              3.5708e-05,
              3.4584e-05,
              3.4e-05,
              3.5916e-05,
              3.2792e-05,
              3.3292e-05,
              3.1666e-05
            ],
            "raw_process_cpu_seconds": [
              5.1e-05,
              4.5e-05,
              4e-05,
              4.3e-05,
              4.7e-05,
              3.7e-05,
              4e-05,
              4.2e-05,
              3.8e-05,
              3.7e-05,
              3.1e-05,
              3.8e-05,
              3.7e-05,
              3.8e-05,
              3.5e-05,
              3.4e-05,
              3.9e-05,
              4e-05,
              3.9e-05,
              3.6e-05,
              3.5e-05,
              4.6e-05,
              3.6e-05,
              3.3e-05,
              3.7e-05,
              3.6e-05,
              3.6e-05,
              3.9e-05,
              3.5e-05,
              3.6e-05,
              3.4e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006190667001646943,
            "speedup_vs_optimized_serial": 0.022415305804771742,
            "speedup_vs_original_serial": 0.02003792488608383
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.333e-06,
            "mad_seconds": 1.2499999999999999e-06,
            "median_process_cpu_seconds": 1.5e-05,
            "median_process_cpu_wall_ratio": 1.92,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 136369.83499249967,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              9.917e-06,
              8.417e-06,
              7.917e-06,
              8.292e-06,
              8.917e-06,
              5.75e-06,
              1.1291e-05,
              6.625e-06,
              1.1084e-05,
              7.167e-06,
              1.5083e-05,
              7.916e-06,
              1.2625e-05,
              5.208e-06,
              6.25e-06,
              6.333e-06,
              5.667e-06,
              5.291e-06,
              7.791e-06,
              1.0917e-05,
              6.792e-06,
              1.275e-05,
              6.75e-06,
              6.083e-06,
              7.5e-06,
              5.416e-06,
              4.083e-06,
              7.333e-06,
              3.584e-06,
              7.5e-06,
              6.666e-06
            ],
            "raw_process_cpu_seconds": [
              1.5e-05,
              1.6e-05,
              1.6e-05,
              1.6e-05,
              1.3e-05,
              1.2e-05,
              1.7e-05,
              1.5e-05,
              1.7e-05,
              1.6e-05,
              1.9e-05,
              1.5e-05,
              2.2e-05,
              1.1e-05,
              1.2e-05,
              1.5e-05,
              1.1e-05,
              1.1e-05,
              1.2e-05,
              2e-05,
              1.3e-05,
              2.2e-05,
              1.4e-05,
              1.2e-05,
              1.6e-05,
              1e-05,
              9e-06,
              1.6e-05,
              1e-05,
              1.3e-05,
              1.2e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005877750001673121,
            "speedup_vs_optimized_serial": 0.10800490931405973,
            "speedup_vs_original_serial": 0.09654984317468976
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 5.3083e-05,
            "mad_seconds": 9.291000000000003e-06,
            "median_process_cpu_seconds": 7.6e-05,
            "median_process_cpu_wall_ratio": 1.4325522979708778,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 18838.422847239228,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              5.7166e-05,
              4.2042e-05,
              5.4125e-05,
              4.2291e-05,
              4.9625e-05,
              4.25e-05,
              5.7209e-05,
              5.0958e-05,
              4.5875e-05,
              4.3792e-05,
              3.8125e-05,
              4.0417e-05,
              5.3333e-05,
              4.4083e-05,
              4.7417e-05,
              4.75e-05,
              6.575e-05,
              0.000119042,
              6.9542e-05,
              8.675e-05,
              0.000120791,
              6.7208e-05,
              6.0792e-05,
              5.775e-05,
              7.4959e-05,
              6.7291e-05,
              6.3375e-05,
              5.3083e-05,
              4.7625e-05,
              4.5125e-05,
              3.7583e-05
            ],
            "raw_process_cpu_seconds": [
              8.9e-05,
              6.1e-05,
              7.6e-05,
              6.1e-05,
              7.6e-05,
              6.5e-05,
              7.9e-05,
              7.3e-05,
              6.4e-05,
              6.3e-05,
              5.6e-05,
              5.8e-05,
              7.1e-05,
              6.8e-05,
              6.6e-05,
              6.6e-05,
              9.6e-05,
              0.000139,
              8.8e-05,
              0.000107,
              0.000144,
              9.5e-05,
              8.5e-05,
              8.2e-05,
              0.000101,
              0.000104,
              9.5e-05,
              8e-05,
              6.7e-05,
              6.5e-05,
              5.6e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.0077269579996936955,
            "speedup_vs_optimized_serial": 0.014920030895013469,
            "speedup_vs_original_serial": 0.013337603375845376
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 2.1e-05,
            "mad_seconds": 4.791000000000002e-06,
            "median_process_cpu_seconds": 5e-05,
            "median_process_cpu_wall_ratio": 2.7735325748570783,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 47619.04761904762,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              2.2084e-05,
              6.75e-06,
              2.1e-05,
              2.0417e-05,
              2.1041e-05,
              3.4875e-05,
              1.8416e-05,
              2.0834e-05,
              2.6083e-05,
              2.4417e-05,
              2.9375e-05,
              1.4209e-05,
              1.9125e-05,
              2.5291e-05,
              1.4459e-05,
              2.2e-05,
              1.0042e-05,
              3.1917e-05,
              9.417e-06,
              1.0708e-05,
              7.667e-06,
              2.0917e-05,
              2.2959e-05,
              1.7667e-05,
              2.4125e-05,
              3.5875e-05,
              5.875e-06,
              2.9209e-05,
              2.5791e-05,
              8.167e-06,
              2.2375e-05
            ],
            "raw_process_cpu_seconds": [
              5.9e-05,
              2.3e-05,
              5.9e-05,
              3.6e-05,
              4.9e-05,
              0.000103,
              5e-05,
              6.5e-05,
              7.2e-05,
              4.5e-05,
              5.6e-05,
              4.5e-05,
              4.4e-05,
              8e-05,
              4.9e-05,
              5.6e-05,
              3.1e-05,
              9.1e-05,
              2.7e-05,
              4.8e-05,
              2.7e-05,
              5.4e-05,
              4.8e-05,
              4.9e-05,
              7.3e-05,
              9.3e-05,
              1.6e-05,
              7.3e-05,
              5.6e-05,
              2.6e-05,
              7.1e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006906707996677142,
            "speedup_vs_optimized_serial": 0.037714285714285714,
            "speedup_vs_original_serial": 0.03371428571428572
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 8.4959e-05,
            "mad_seconds": 1.0374000000000006e-05,
            "median_process_cpu_seconds": 0.000156,
            "median_process_cpu_wall_ratio": 1.8305941845764855,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 11770.38336138608,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000164167,
              9.3334e-05,
              9.9875e-05,
              7.7333e-05,
              9.5333e-05,
              7.7625e-05,
              0.000106375,
              7.9875e-05,
              9.4125e-05,
              0.000116958,
              8.7875e-05,
              8.1917e-05,
              8.4959e-05,
              0.000102167,
              6.7584e-05,
              7.8625e-05,
              9.0417e-05,
              8.2542e-05,
              7.4375e-05,
              0.00010025,
              0.000107583,
              7.2208e-05,
              8.8333e-05,
              7.7417e-05,
              6.4417e-05,
              7.4083e-05,
              5.8875e-05,
              6.375e-05,
              9.8875e-05,
              7.8167e-05,
              8.8417e-05
            ],
            "raw_process_cpu_seconds": [
              0.000324,
              0.000171,
              0.000201,
              0.000145,
              0.000167,
              0.000138,
              0.000187,
              0.000158,
              0.000197,
              0.000205,
              0.00018,
              0.000144,
              0.000136,
              0.000215,
              0.000121,
              0.000134,
              0.000151,
              0.000162,
              0.000133,
              0.000186,
              0.000235,
              0.000126,
              0.000179,
              0.000153,
              0.000106,
              0.000146,
              0.000102,
              0.000109,
              0.000181,
              0.000147,
              0.000156
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.00918608399661025,
            "speedup_vs_optimized_serial": 0.009322143622217775,
            "speedup_vs_original_serial": 0.008333431419861346
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 9.59e-07,
            "mad_seconds": 4.0999999999999904e-08,
            "median_process_cpu_seconds": 1e-06,
            "median_process_cpu_wall_ratio": 1.04384133611691,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1042752.8675703858,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1e-06,
              2e-06,
              1.042e-06,
              1.208e-06,
              1.125e-06,
              2.167e-06,
              1e-06,
              9.59e-07,
              9.58e-07,
              1e-06,
              9.59e-07,
              9.58e-07,
              9.59e-07,
              9.58e-07,
              1e-06,
              9.17e-07,
              9.17e-07,
              9.58e-07,
              9.58e-07,
              9.59e-07,
              9.58e-07,
              1.166e-06,
              9.17e-07,
              1e-06,
              9.59e-07,
              9.58e-07,
              9.16e-07,
              9.17e-07,
              9.59e-07,
              9.17e-07,
              9.16e-07
            ],
            "raw_process_cpu_seconds": [
              2e-06,
              2e-06,
              1e-06,
              2e-06,
              1e-06,
              3e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              2e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06,
              1e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005455167003674433
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 5.375e-06,
            "mad_seconds": 9.160000000000002e-07,
            "median_process_cpu_seconds": 8e-06,
            "median_process_cpu_wall_ratio": 1.3715539706487452,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 186046.51162790696,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              4.625e-06,
              5.375e-06,
              5.75e-06,
              5.791e-06,
              4.334e-06,
              7.75e-06,
              6.042e-06,
              7.25e-06,
              8.75e-06,
              4.584e-06,
              6.667e-06,
              4.416e-06,
              5.375e-06,
              4.25e-06,
              5.292e-06,
              4.583e-06,
              4.916e-06,
              6.583e-06,
              7.375e-06,
              9.334e-06,
              6e-06,
              4.625e-06,
              1.0167e-05,
              7.291e-06,
              5.542e-06,
              4.208e-06,
              4.375e-06,
              4.417e-06,
              4.583e-06,
              5.083e-06,
              4.459e-06
            ],
            "raw_process_cpu_seconds": [
              8e-06,
              9e-06,
              8e-06,
              8e-06,
              5e-06,
              8e-06,
              1.1e-05,
              9e-06,
              1e-05,
              7e-06,
              9e-06,
              6e-06,
              7e-06,
              6e-06,
              9e-06,
              8e-06,
              5e-06,
              1.2e-05,
              9e-06,
              1.3e-05,
              7e-06,
              6e-06,
              1.3e-05,
              1e-05,
              7e-06,
              5e-06,
              7e-06,
              7e-06,
              7e-06,
              8e-06,
              6e-06
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.005451124998216983,
            "speedup_vs_optimized_serial": 0.1784186046511628
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 3.6042e-05,
            "mad_seconds": 3.7499999999999963e-06,
            "median_process_cpu_seconds": 3.8e-05,
            "median_process_cpu_wall_ratio": 1.0801412087043576,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 27745.40813495367,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              6.25e-05,
              4.4291e-05,
              4.875e-05,
              3.6542e-05,
              3.275e-05,
              4.075e-05,
              4.45e-05,
              3.6042e-05,
              4.2208e-05,
              3.5084e-05,
              4.4333e-05,
              3.5666e-05,
              3.0167e-05,
              3.1625e-05,
              3.7958e-05,
              2.7875e-05,
              3.9333e-05,
              3.5041e-05,
              3.85e-05,
              2.7833e-05,
              3.2292e-05,
              3.7958e-05,
              3.4625e-05,
              3.675e-05,
              3.5083e-05,
              3.15e-05,
              3.2166e-05,
              3.8292e-05,
              2.9917e-05,
              3.825e-05,
              3.0875e-05
            ],
            "raw_process_cpu_seconds": [
              6.7e-05,
              4.7e-05,
              5.5e-05,
              3.9e-05,
              3.7e-05,
              4.4e-05,
              4.7e-05,
              3.6e-05,
              5.1e-05,
              3.8e-05,
              4.8e-05,
              3.8e-05,
              3.4e-05,
              3.4e-05,
              4.1e-05,
              3.2e-05,
              4.2e-05,
              3.8e-05,
              4.1e-05,
              3.2e-05,
              3.6e-05,
              4e-05,
              3.7e-05,
              4e-05,
              3.8e-05,
              3.4e-05,
              3.5e-05,
              4.3e-05,
              3.4e-05,
              4.1e-05,
              3.2e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006706625004881062,
            "speedup_vs_optimized_serial": 0.026607846401420568
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 8.291e-06,
            "mad_seconds": 2.1240000000000003e-06,
            "median_process_cpu_seconds": 1.5e-05,
            "median_process_cpu_wall_ratio": 1.8091906886985887,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 120612.71257990591,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1.425e-05,
              1.7417e-05,
              1.0083e-05,
              7.792e-06,
              7.084e-06,
              7.25e-06,
              1.4125e-05,
              1.3833e-05,
              1.175e-05,
              4.917e-06,
              7.708e-06,
              1.6042e-05,
              6.375e-06,
              8.375e-06,
              1.6792e-05,
              1.4875e-05,
              1.1917e-05,
              1.8167e-05,
              9.541e-06,
              6.292e-06,
              7.416e-06,
              6.167e-06,
              2.667e-06,
              1.1625e-05,
              6.75e-06,
              1.025e-05,
              7.083e-06,
              5.75e-06,
              6.75e-06,
              5.791e-06,
              8.291e-06
            ],
            "raw_process_cpu_seconds": [
              2.2e-05,
              2.9e-05,
              1.2e-05,
              1.4e-05,
              1.4e-05,
              1.5e-05,
              3e-05,
              2.4e-05,
              1.5e-05,
              1.2e-05,
              1.6e-05,
              2.4e-05,
              1.1e-05,
              1.7e-05,
              2.2e-05,
              2e-05,
              2.3e-05,
              2.9e-05,
              1.2e-05,
              1.1e-05,
              1.9e-05,
              1.4e-05,
              6e-06,
              1.5e-05,
              1.2e-05,
              2e-05,
              1.3e-05,
              1.1e-05,
              1.3e-05,
              1.1e-05,
              1.5e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006212209002114832,
            "speedup_vs_optimized_serial": 0.11566759136412977
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 4.875e-05,
            "mad_seconds": 5.583999999999999e-06,
            "median_process_cpu_seconds": 7.3e-05,
            "median_process_cpu_wall_ratio": 1.4861825192802056,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 20512.82051282051,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              5.5917e-05,
              6.7917e-05,
              7.825e-05,
              4.525e-05,
              4.3166e-05,
              4.5875e-05,
              5.8334e-05,
              5.975e-05,
              5.5667e-05,
              4.4875e-05,
              4.1917e-05,
              3.9958e-05,
              4.4166e-05,
              6.2958e-05,
              7.4291e-05,
              6.45e-05,
              5.0625e-05,
              4.8292e-05,
              4.875e-05,
              4.8584e-05,
              6.3625e-05,
              5.3042e-05,
              3.7917e-05,
              4.6125e-05,
              4.3792e-05,
              4.5875e-05,
              4.5375e-05,
              3.875e-05,
              4.9792e-05,
              4.9208e-05,
              5.5042e-05
            ],
            "raw_process_cpu_seconds": [
              8.9e-05,
              9.9e-05,
              0.000115,
              6.8e-05,
              6.1e-05,
              6.4e-05,
              8.8e-05,
              8.3e-05,
              8.3e-05,
              6.4e-05,
              6.1e-05,
              6.2e-05,
              7e-05,
              9.6e-05,
              0.000108,
              9.1e-05,
              7.8e-05,
              7.2e-05,
              7.3e-05,
              6.9e-05,
              9.5e-05,
              7.9e-05,
              5.6e-05,
              7.2e-05,
              6.3e-05,
              6.5e-05,
              6.2e-05,
              6.1e-05,
              7.4e-05,
              7.3e-05,
              8.2e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.007466249997378327,
            "speedup_vs_optimized_serial": 0.019671794871794872
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 1.7125e-05,
            "mad_seconds": 6.874999999999999e-06,
            "median_process_cpu_seconds": 5.1e-05,
            "median_process_cpu_wall_ratio": 2.9281736234713214,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 58394.16058394161,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              1.875e-05,
              4.375e-05,
              1.7666e-05,
              1.0333e-05,
              4.0709e-05,
              2.4125e-05,
              4.667e-06,
              7.667e-06,
              2.275e-05,
              3.075e-05,
              1.8959e-05,
              2.8917e-05,
              3.0417e-05,
              3.75e-05,
              6.208e-06,
              1.1542e-05,
              1.4e-05,
              1.7125e-05,
              6.5e-06,
              1.7375e-05,
              4.0875e-05,
              1.45e-05,
              1.025e-05,
              4.334e-06,
              1.7e-05,
              1.1333e-05,
              1.7417e-05,
              1.7541e-05,
              7.708e-06,
              4.625e-06,
              1.2708e-05
            ],
            "raw_process_cpu_seconds": [
              6.4e-05,
              8.5e-05,
              4e-05,
              2.9e-05,
              0.000112,
              6.9e-05,
              1.9e-05,
              2.6e-05,
              5.6e-05,
              9e-05,
              5e-05,
              5.6e-05,
              7.7e-05,
              7.4e-05,
              1.2e-05,
              5.3e-05,
              5.3e-05,
              3.4e-05,
              2.2e-05,
              5.1e-05,
              8.4e-05,
              3.7e-05,
              3.9e-05,
              2e-05,
              6.1e-05,
              3.9e-05,
              5.1e-05,
              5.4e-05,
              3e-05,
              1.7e-05,
              4.1e-05
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.006091333998483606,
            "speedup_vs_optimized_serial": 0.056
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 32,
            "q": 1,
            "d": 4,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 3,
            "median_seconds": 7.9875e-05,
            "mad_seconds": 9.95899999999999e-06,
            "median_process_cpu_seconds": 0.000153,
            "median_process_cpu_wall_ratio": 1.829937782115408,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 12519.561815336463,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              7.9875e-05,
              8.9834e-05,
              8.375e-05,
              6.6584e-05,
              8.8417e-05,
              6.2083e-05,
              9.0167e-05,
              5.7958e-05,
              7.7041e-05,
              8.2916e-05,
              6.9625e-05,
              6.3041e-05,
              6.175e-05,
              6.5125e-05,
              7.1584e-05,
              9.3792e-05,
              0.000108041,
              7.2375e-05,
              8.6666e-05,
              0.000100333,
              8.275e-05,
              7.3667e-05,
              8.6834e-05,
              7.1542e-05,
              6.6792e-05,
              0.000105125,
              8.6834e-05,
              0.000105333,
              7.1e-05,
              7.5042e-05,
              9.275e-05
            ],
            "raw_process_cpu_seconds": [
              0.000168,
              0.000163,
              0.000167,
              0.00012,
              0.000164,
              0.000108,
              0.000165,
              9.9e-05,
              0.000128,
              0.00018,
              0.000121,
              0.00011,
              0.000111,
              0.000119,
              0.000144,
              0.000208,
              0.000186,
              0.000134,
              0.00017,
              0.000169,
              0.00017,
              0.000139,
              0.000153,
              0.000118,
              0.00012,
              0.000206,
              0.000166,
              0.00019,
              0.000133,
              0.00014,
              0.000182
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "32",
              "--q",
              "1",
              "--d",
              "4",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "31",
              "--warmups",
              "3",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.008090292001725174,
            "speedup_vs_optimized_serial": 0.012006259780907668
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "integrated-wide",
      "kernel": "final-borrowed-stop-token-with-concurrent-source-integration",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:35:15Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "71cdb79f8d322e636bf6a595bbd4bf92f5593a66e62ade3ae7c12cbafd776052"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.169096416,
            "mad_seconds": 0.002105709000000011,
            "median_process_cpu_seconds": 0.168796,
            "median_process_cpu_wall_ratio": 0.9984497026298198,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 378.48229734212697,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.165693333,
              0.165601083,
              0.168496584,
              0.169096416,
              0.173195125,
              0.171202125,
              0.170182834
            ],
            "raw_process_cpu_seconds": [
              0.165524,
              0.165251,
              0.168252,
              0.168796,
              0.173123,
              0.170734,
              0.169919
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.1956913329995587
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.170148541,
            "mad_seconds": 0.0032807489999999995,
            "median_process_cpu_seconds": 0.170081,
            "median_process_cpu_wall_ratio": 0.999603046845991,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 376.1419264829312,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.1828055,
              0.178059583,
              0.171751875,
              0.169536708,
              0.170148541,
              0.165533125,
              0.166867792
            ],
            "raw_process_cpu_seconds": [
              0.182476,
              0.177912,
              0.171641,
              0.169482,
              0.170081,
              0.165481,
              0.166828
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.2413449159939773,
            "speedup_vs_original_serial": 0.9938164324312367
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.164497625,
            "mad_seconds": 0.0004241249999999974,
            "median_process_cpu_seconds": 0.164322,
            "median_process_cpu_wall_ratio": 0.9995129358076495,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 389.0633679361632,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.1640735,
              0.164735917,
              0.164110334,
              0.164052167,
              0.166362916,
              0.164497625,
              0.166045875
            ],
            "raw_process_cpu_seconds": [
              0.163995,
              0.164634,
              0.163948,
              0.163989,
              0.166287,
              0.164322,
              0.165965
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.84556899999734,
            "speedup_vs_optimized_serial": 1.0343525689200679,
            "speedup_vs_original_serial": 1.0279565799202268
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.165110083,
            "mad_seconds": 0.0013410419999999867,
            "median_process_cpu_seconds": 0.16504,
            "median_process_cpu_wall_ratio": 0.999575537733816,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 387.6201794411308,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.165110083,
              0.163936458,
              0.168086875,
              0.16537225,
              0.166964167,
              0.163769041,
              0.160335084
            ],
            "raw_process_cpu_seconds": [
              0.16504,
              0.16387,
              0.168072,
              0.165294,
              0.166879,
              0.163655,
              0.160302
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.1842860830001882,
            "speedup_vs_optimized_serial": 1.0305157499072906,
            "speedup_vs_original_serial": 1.0241434861370642
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.087142292,
            "mad_seconds": 0.0028195419999999943,
            "median_process_cpu_seconds": 0.172552,
            "median_process_cpu_wall_ratio": 1.9860063576497504,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 734.4309924737807,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.086157125,
              0.084843917,
              0.090423917,
              0.094815917,
              0.090886416,
              0.087142292,
              0.08432275
            ],
            "raw_process_cpu_seconds": [
              0.171885,
              0.169458,
              0.179059,
              0.187382,
              0.180501,
              0.172552,
              0.167838
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.903386625002895,
            "speedup_vs_optimized_serial": 1.9525369036655589,
            "speedup_vs_original_serial": 1.940463259791239
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.085178625,
            "mad_seconds": 0.0005613339999999967,
            "median_process_cpu_seconds": 0.16957,
            "median_process_cpu_wall_ratio": 1.9927759586698683,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 751.3622108833056,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.085904416,
              0.08589375,
              0.085033542,
              0.0868135,
              0.084617291,
              0.084654292,
              0.085178625
            ],
            "raw_process_cpu_seconds": [
              0.170877,
              0.171167,
              0.16957,
              0.172341,
              0.16901,
              0.168898,
              0.168794
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.5371424160039169,
            "speedup_vs_optimized_serial": 1.9975497491301368,
            "speedup_vs_original_serial": 1.9851977652844244
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.043960958,
            "mad_seconds": 0.00027066699999999527,
            "median_process_cpu_seconds": 0.173947,
            "median_process_cpu_wall_ratio": 3.9724648177896786,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1455.8372454030687,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.04370775,
              0.044231625,
              0.044156042,
              0.044293167,
              0.043638958,
              0.043495917,
              0.043960958
            ],
            "raw_process_cpu_seconds": [
              0.173888,
              0.175236,
              0.173947,
              0.174383,
              0.173654,
              0.172786,
              0.174857
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.386052916997869,
            "speedup_vs_optimized_serial": 3.8704466131061106,
            "speedup_vs_original_serial": 3.846513444952678
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.043865833,
            "mad_seconds": 0.0001633340000000011,
            "median_process_cpu_seconds": 0.17426,
            "median_process_cpu_wall_ratio": 3.9577995848269847,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1458.9942928930586,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.043865833,
              0.043803125,
              0.044029167,
              0.043574042,
              0.044185916,
              0.044419042,
              0.043861416
            ],
            "raw_process_cpu_seconds": [
              0.171943,
              0.172596,
              0.174472,
              0.173229,
              0.174879,
              0.17573,
              0.17426
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.2178190420017927,
            "speedup_vs_optimized_serial": 3.8788398478606343,
            "speedup_vs_original_serial": 3.8548547795729764
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.169725042,
            "mad_seconds": 0.0032089999999999896,
            "median_process_cpu_seconds": 0.169717,
            "median_process_cpu_wall_ratio": 0.9993284442887646,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 377.080478200736,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.165457625,
              0.179694292,
              0.168079875,
              0.166516042,
              0.169725042,
              0.172162708,
              0.173903584
            ],
            "raw_process_cpu_seconds": [
              0.16533,
              0.178438,
              0.167967,
              0.166433,
              0.169717,
              0.172134,
              0.173723
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.223031292000087
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.164080542,
            "mad_seconds": 0.0005993330000000074,
            "median_process_cpu_seconds": 0.163946,
            "median_process_cpu_wall_ratio": 0.9993874479197898,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 390.0523439275329,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.162863,
              0.163953875,
              0.164679875,
              0.16417125,
              0.16162375,
              0.164080542,
              0.164723584
            ],
            "raw_process_cpu_seconds": [
              0.162835,
              0.163891,
              0.164579,
              0.164036,
              0.161554,
              0.163946,
              0.164589
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.8394713749949005,
            "speedup_vs_optimized_serial": 1.0344007883640463
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.164149417,
            "mad_seconds": 0.00037483300000001885,
            "median_process_cpu_seconds": 0.164105,
            "median_process_cpu_wall_ratio": 0.9996270582838649,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 389.888682943053,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.164149417,
              0.164039292,
              0.16452425,
              0.164457333,
              0.163699209,
              0.16495625,
              0.163533792
            ],
            "raw_process_cpu_seconds": [
              0.164105,
              0.163979,
              0.164413,
              0.164396,
              0.163626,
              0.1649,
              0.16346
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 2.1558086670047487,
            "speedup_vs_optimized_serial": 1.033966766997412
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.085464125,
            "mad_seconds": 0.0006180840000000049,
            "median_process_cpu_seconds": 0.169496,
            "median_process_cpu_wall_ratio": 1.9904401959658033,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 748.8522230819071,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.086890167,
              0.084846041,
              0.083296583,
              0.085464125,
              0.085958875,
              0.085609208,
              0.0845105
            ],
            "raw_process_cpu_seconds": [
              0.173046,
              0.169496,
              0.166326,
              0.168931,
              0.171096,
              0.170324,
              0.167865
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.8678842910012463,
            "speedup_vs_optimized_serial": 1.985921484599532
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.085049666,
            "mad_seconds": 0.000512665999999995,
            "median_process_cpu_seconds": 0.168767,
            "median_process_cpu_wall_ratio": 1.9910151039710402,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 752.5014854261744,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.083616708,
              0.085130417,
              0.084537,
              0.085049666,
              0.08453075,
              0.085984,
              0.085516041
            ],
            "raw_process_cpu_seconds": [
              0.166793,
              0.168767,
              0.167533,
              0.169214,
              0.168302,
              0.171283,
              0.170587
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.5316009579983074,
            "speedup_vs_optimized_serial": 1.995599159672185
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.044048791,
            "mad_seconds": 0.00014679199999999976,
            "median_process_cpu_seconds": 0.174528,
            "median_process_cpu_wall_ratio": 3.969799878581463,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1452.934315495742,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.0437885,
              0.044048791,
              0.044195583,
              0.044112,
              0.043802208,
              0.044165958,
              0.043856167
            ],
            "raw_process_cpu_seconds": [
              0.174135,
              0.174466,
              0.175137,
              0.174596,
              0.173886,
              0.175697,
              0.174528
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.3814709590005805,
            "speedup_vs_optimized_serial": 3.8531146518868136
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 4096,
            "q": 64,
            "d": 64,
            "k": 5,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.04425425,
            "mad_seconds": 0.00027695800000000076,
            "median_process_cpu_seconds": 0.174855,
            "median_process_cpu_wall_ratio": 3.949881469671708,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1446.188784128078,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.043977292,
              0.044268417,
              0.043758,
              0.04447225,
              0.043918125,
              0.04425425,
              0.045572583
            ],
            "raw_process_cpu_seconds": [
              0.173549,
              0.174855,
              0.173644,
              0.175158,
              0.174155,
              0.175341,
              0.177498
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "4096",
              "--q",
              "64",
              "--d",
              "64",
              "--k",
              "5",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 1.2477617919939803,
            "speedup_vs_optimized_serial": 3.8352258144697964
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "integrated-batch-k32",
      "kernel": "final-borrowed-stop-token-with-concurrent-source-integration",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:35:44Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "71cdb79f8d322e636bf6a595bbd4bf92f5593a66e62ade3ae7c12cbafd776052"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.055395209,
            "mad_seconds": 0.00024416599999999705,
            "median_process_cpu_seconds": 0.055392,
            "median_process_cpu_wall_ratio": 0.9996842950308037,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4621.3382821608275,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.053385083,
              0.05409575,
              0.055639375,
              0.055765,
              0.055395209,
              0.0551915,
              0.0554315
            ],
            "raw_process_cpu_seconds": [
              0.053297,
              0.054036,
              0.055599,
              0.05575,
              0.055392,
              0.055175,
              0.055414
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7392157919966849
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.05553975,
            "mad_seconds": 0.0007877919999999955,
            "median_process_cpu_seconds": 0.055448,
            "median_process_cpu_wall_ratio": 0.998714555408374,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4609.311349078813,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.054057417,
              0.05553975,
              0.055881833,
              0.056202125,
              0.054428292,
              0.054751958,
              0.069546459
            ],
            "raw_process_cpu_seconds": [
              0.053982,
              0.055448,
              0.05581,
              0.056184,
              0.054397,
              0.054726,
              0.067876
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.760043707996374,
            "speedup_vs_original_serial": 0.9973975215948938
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.053664333,
            "mad_seconds": 0.00018900000000000167,
            "median_process_cpu_seconds": 0.053638,
            "median_process_cpu_wall_ratio": 0.9986205511512007,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4770.393773458434,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.053853333,
              0.053601959,
              0.053664833,
              0.053385375,
              0.053382584,
              0.053664333,
              0.053856292
            ],
            "raw_process_cpu_seconds": [
              0.053707,
              0.0535,
              0.053638,
              0.053325,
              0.053276,
              0.05364,
              0.053782
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9569803749982384,
            "speedup_vs_optimized_serial": 1.03494717804468,
            "speedup_vs_original_serial": 1.032253750363393
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.055653458,
            "mad_seconds": 9.804199999999957e-05,
            "median_process_cpu_seconds": 0.055604,
            "median_process_cpu_wall_ratio": 0.9996961496876715,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4599.893864636408,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.053866958,
              0.0555835,
              0.0556885,
              0.055653458,
              0.054852667,
              0.0557515,
              0.055807625
            ],
            "raw_process_cpu_seconds": [
              0.053852,
              0.055586,
              0.055604,
              0.055633,
              0.054836,
              0.055742,
              0.05579
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7428311250041588,
            "speedup_vs_optimized_serial": 0.9979568565173433,
            "speedup_vs_original_serial": 0.9953596953490293
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028392375,
            "mad_seconds": 0.00020645799999999964,
            "median_process_cpu_seconds": 0.0561,
            "median_process_cpu_wall_ratio": 1.9884898944666962,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 9016.50531172542,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.028416625,
              0.028505625,
              0.027982541,
              0.028185917,
              0.028392375,
              0.028880917,
              0.027725209
            ],
            "raw_process_cpu_seconds": [
              0.05652,
              0.056716,
              0.055643,
              0.055945,
              0.0561,
              0.05685,
              0.055206
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.6443981659976998,
            "speedup_vs_optimized_serial": 1.9561501987769603,
            "speedup_vs_original_serial": 1.951059360127499
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.027977292,
            "mad_seconds": 0.00018066599999999947,
            "median_process_cpu_seconds": 0.055436,
            "median_process_cpu_wall_ratio": 1.987395534860873,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 9150.278018330007,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.027864083,
              0.028332375,
              0.028648,
              0.0276775,
              0.027890083,
              0.028157958,
              0.027977292
            ],
            "raw_process_cpu_seconds": [
              0.055329,
              0.055854,
              0.057022,
              0.055213,
              0.055436,
              0.055961,
              0.0553
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.5303690419968916,
            "speedup_vs_optimized_serial": 1.9851724748771253,
            "speedup_vs_original_serial": 1.9800061063808463
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014390666,
            "mad_seconds": 5.462600000000033e-05,
            "median_process_cpu_seconds": 0.056939,
            "median_process_cpu_wall_ratio": 3.969314790999076,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17789.308708853365,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014428041,
              0.014445292,
              0.014422,
              0.014290625,
              0.014236292,
              0.014390666,
              0.014312042
            ],
            "raw_process_cpu_seconds": [
              0.057358,
              0.05709,
              0.057151,
              0.05675,
              0.056562,
              0.056939,
              0.056809
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.47675566699763294,
            "speedup_vs_optimized_serial": 3.8594287436036665,
            "speedup_vs_original_serial": 3.8493846636423914
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.014489084,
            "mad_seconds": 0.00012204299999999876,
            "median_process_cpu_seconds": 0.057112,
            "median_process_cpu_wall_ratio": 3.930245622602455,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17668.47372822188,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014489084,
              0.014367041,
              0.014489875,
              0.014288375,
              0.014678416,
              0.014654,
              0.014437625
            ],
            "raw_process_cpu_seconds": [
              0.056696,
              0.056466,
              0.057275,
              0.056523,
              0.057387,
              0.057112,
              0.057148
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.4238449580007,
            "speedup_vs_optimized_serial": 3.8332133349492628,
            "speedup_vs_original_serial": 3.823237480022892
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.05598675,
            "mad_seconds": 0.0006938339999999973,
            "median_process_cpu_seconds": 0.055901,
            "median_process_cpu_wall_ratio": 0.9991502236781192,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4572.510460064212,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.057643667,
              0.056680584,
              0.056710792,
              0.05598675,
              0.055652292,
              0.055152834,
              0.055487
            ],
            "raw_process_cpu_seconds": [
              0.057636,
              0.056563,
              0.056689,
              0.055901,
              0.055605,
              0.055101,
              0.055473
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7573179999963031
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.05485425,
            "mad_seconds": 0.0005178330000000023,
            "median_process_cpu_seconds": 0.054781,
            "median_process_cpu_wall_ratio": 0.9993446762674203,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4666.912773394951,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.055799,
              0.05485425,
              0.053089791,
              0.054479292,
              0.054216584,
              0.0550115,
              0.055372083
            ],
            "raw_process_cpu_seconds": [
              0.055797,
              0.054781,
              0.053055,
              0.054411,
              0.054194,
              0.054841,
              0.055354
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.9556001250020927,
            "speedup_vs_optimized_serial": 1.0206456199838663
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.054377083,
            "mad_seconds": 0.00033250099999999894,
            "median_process_cpu_seconds": 0.054261,
            "median_process_cpu_wall_ratio": 0.9995570666114556,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 4707.8656278785675,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.053529334,
              0.054155667,
              0.054746042,
              0.054279042,
              0.054377083,
              0.054709584,
              0.055129833
            ],
            "raw_process_cpu_seconds": [
              0.053512,
              0.054144,
              0.054597,
              0.054255,
              0.054261,
              0.054688,
              0.055101
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.7394241250003688,
            "speedup_vs_optimized_serial": 1.0296019372719938
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.028191333,
            "mad_seconds": 0.000301667000000002,
            "median_process_cpu_seconds": 0.055774,
            "median_process_cpu_wall_ratio": 1.9822983557655383,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 9080.805082895513,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.027967,
              0.028516083,
              0.028493,
              0.028503417,
              0.02799175,
              0.028191333,
              0.027605209
            ],
            "raw_process_cpu_seconds": [
              0.055774,
              0.05583,
              0.056095,
              0.056509,
              0.055488,
              0.055727,
              0.055032
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.6487911660005921,
            "speedup_vs_optimized_serial": 1.985956109276564
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.029611083,
            "mad_seconds": 0.0004217920000000007,
            "median_process_cpu_seconds": 0.058727,
            "median_process_cpu_wall_ratio": 1.9638366687289528,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 8645.4115845746,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.029189291,
              0.029233834,
              0.031150208,
              0.034417584,
              0.037911417,
              0.029611083,
              0.029583875
            ],
            "raw_process_cpu_seconds": [
              0.057323,
              0.057627,
              0.061015,
              0.062926,
              0.068129,
              0.058727,
              0.058599
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.5679034170025261,
            "speedup_vs_optimized_serial": 1.8907363165339142
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.018796875,
            "mad_seconds": 0.0004734579999999995,
            "median_process_cpu_seconds": 0.064004,
            "median_process_cpu_wall_ratio": 3.5981363453420494,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 13619.285120532002,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.016902917,
              0.014533333,
              0.015884,
              0.019270333,
              0.018819916,
              0.018896,
              0.018796875
            ],
            "raw_process_cpu_seconds": [
              0.060819,
              0.057785,
              0.059496,
              0.064004,
              0.068147,
              0.066293,
              0.066501
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.517578708997462,
            "speedup_vs_optimized_serial": 2.978513715710723
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 2048,
            "q": 256,
            "d": 16,
            "k": 32,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.015044083,
            "mad_seconds": 0.0004875000000000001,
            "median_process_cpu_seconds": 0.058964,
            "median_process_cpu_wall_ratio": 3.9194146961300333,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 17016.65698068802,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.014716209,
              0.014556583,
              0.015044083,
              0.018195625,
              0.022053,
              0.018258,
              0.014618375
            ],
            "raw_process_cpu_seconds": [
              0.058133,
              0.057752,
              0.058964,
              0.064783,
              0.072495,
              0.061036,
              0.058046
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "2048",
              "--q",
              "256",
              "--d",
              "16",
              "--k",
              "32",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "7",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.45131304200185696,
            "speedup_vs_optimized_serial": 3.7215129695841216
          }
        ],
        "skipped": []
      }
    },
    {
      "name": "integrated-all-neighbors",
      "kernel": "final-borrowed-stop-token-with-concurrent-source-integration",
      "report": {
        "schema": "aner.knn-native-benchmark/1",
        "timestamp_utc": "2026-10-30T17:35:54Z",
        "machine": {
          "platform": "macOS-26.7.2-arm64-arm-64bit",
          "system": "Darwin",
          "release": "25.6.0",
          "machine": "arm64",
          "processor": "arm",
          "cpu_model": "arm",
          "logical_cpu_count": 12
        },
        "binaries": {
          "baseline": {
            "sha256": "2d89300b4a0f7123072abf9f80e270edeada5d77b633657f7960421d9beddadb"
          },
          "optimized": {
            "sha256": "ae365f382bbe5f9aca910939c8d3c894e439937c2a920243d787c27b07d9533f"
          }
        },
        "sources": {
          "src/ml.cpp": {
            "sha256": "12f9ef7d5175f0d3c29c7aa1c6b814980541d00c4bd3b322e88931ddc12b0e64"
          },
          "include/aner/ml.hpp": {
            "sha256": "f6340833c00df9da7f60d5f73418f7c74ac5f3ec6f9fe9dbdf5f91c4d35f4cdc"
          },
          "src/parallel.cpp": {
            "sha256": "9e6d226b415322aae21578c0fa7471e8b7233a85128816eb8886a503239bf594"
          },
          "CMakeLists.txt": {
            "sha256": "71cdb79f8d322e636bf6a595bbd4bf92f5593a66e62ade3ae7c12cbafd776052"
          }
        },
        "baseline_sources": {
          "src/ml.cpp": {
            "sha256": "5e6be6bb4338ebdd196a7eef2fc37ea4e773952379bd44fe70ad6e382b80db61"
          },
          "include/aner/ml.hpp": {
            "sha256": "51a82c4e2a7225223f3ec7c86668331cb0e322847bf10c8fef542c1ebbe1677f"
          },
          "CMakeLists.txt": {
            "sha256": "cb4dcffda5cb035ace77357881e1689ec8199d141127cc7f2a22c3ebd48e99fe"
          }
        },
        "driver": {
          "sha256": "2760f5d1c47100c4e3348e4638332907a3562549e69f5d790ecd25b97cf1a261"
        },
        "benchmark_source": {
          "sha256": "f8d51d8f0c2c8af00635faa57094f761af28b8d9f86e6376523d073d13226988"
        },
        "notes": [
          "Cases execute sequentially; do not run builds or tests concurrently.",
          "Original baseline links the preserved pre-change native ml.cpp and uses uniform serial KNN.",
          "Timed regions measure predict() after fit; cold cases include pool creation/destruction.",
          "Complete subprocess wall time also includes independent correctness verification and is not the kernel timer.",
          "Every raw sample verifies exact predictions after stopping its timer; all readouts are checked before timing.",
          "Distance weighting is an optional changed classifier and has no pre-change weighted speed baseline.",
          "CPU slots and worker counts do not measure instantaneous hardware utilization; RSS is not measured."
        ],
        "records": [
          {
            "flavor": "original",
            "mode": "serial",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.001974416,
            "mad_seconds": 3.34990000000002e-05,
            "median_process_cpu_seconds": 0.001973,
            "median_process_cpu_wall_ratio": 1.0002119351508931,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 129658.59271804927,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.00195225,
              0.001967583,
              0.001953792,
              0.00209525,
              0.001840458,
              0.001940917,
              0.0019255,
              0.001974416,
              0.001983792,
              0.002076375,
              0.002020833,
              0.002007417,
              0.001904083,
              0.001979542,
              0.002041125
            ],
            "raw_process_cpu_seconds": [
              0.001955,
              0.001968,
              0.001954,
              0.002075,
              0.001841,
              0.001941,
              0.001926,
              0.001976,
              0.001979,
              0.002077,
              0.002021,
              0.002008,
              0.001886,
              0.001973,
              0.002042
            ],
            "command": [
              "aner_knn_baseline",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.04712629099958576
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000194791,
            "mad_seconds": 4.50099999999999e-06,
            "median_process_cpu_seconds": 0.000195,
            "median_process_cpu_wall_ratio": 1.0018691588785047,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1314229.0968268553,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000192334,
              0.000191959,
              0.000188209,
              0.000192458,
              0.000191458,
              0.00018875,
              0.000230167,
              0.000194791,
              0.000205417,
              0.000200625,
              0.000194834,
              0.000194292,
              0.000202458,
              0.000199292,
              0.000224167
            ],
            "raw_process_cpu_seconds": [
              0.000194,
              0.000192,
              0.000188,
              0.000193,
              0.000192,
              0.000188,
              0.000222,
              0.000195,
              0.000206,
              0.000201,
              0.000194,
              0.000195,
              0.000203,
              0.0002,
              0.000217
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.01597720800054958,
            "speedup_vs_original_serial": 10.136074048595674
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.00020475,
            "mad_seconds": 9.999999999999972e-07,
            "median_process_cpu_seconds": 0.00021,
            "median_process_cpu_wall_ratio": 1.0238393977415308,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1250305.2503052503,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000204333,
              0.000195834,
              0.00019925,
              0.000216541,
              0.000204125,
              0.0002025,
              0.000205,
              0.000204875,
              0.000205708,
              0.000205333,
              0.00021875,
              0.000209083,
              0.000199625,
              0.00020375,
              0.00020475
            ],
            "raw_process_cpu_seconds": [
              0.00021,
              0.0002,
              0.000204,
              0.000221,
              0.000208,
              0.000207,
              0.000211,
              0.00021,
              0.000209,
              0.000211,
              0.000223,
              0.000215,
              0.000205,
              0.000208,
              0.00021
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.020581292003043927,
            "speedup_vs_optimized_serial": 0.9513601953601954,
            "speedup_vs_original_serial": 9.643057387057388
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000247708,
            "mad_seconds": 1.3291000000000019e-05,
            "median_process_cpu_seconds": 0.000253,
            "median_process_cpu_wall_ratio": 1.0213638639042744,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1033474.8978636136,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000272917,
              0.00028275,
              0.000268875,
              0.000258333,
              0.000239541,
              0.000245208,
              0.000265875,
              0.000288041,
              0.000251875,
              0.000247708,
              0.000228958,
              0.0002475,
              0.0002295,
              0.000234417,
              0.000243208
            ],
            "raw_process_cpu_seconds": [
              0.000292,
              0.00029,
              0.000278,
              0.000264,
              0.000243,
              0.000248,
              0.000269,
              0.000302,
              0.000256,
              0.000253,
              0.000238,
              0.00025,
              0.000232,
              0.000241,
              0.000247
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.01702941599796759,
            "speedup_vs_optimized_serial": 0.7863734719912154,
            "speedup_vs_original_serial": 7.970739741954236
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000110125,
            "mad_seconds": 2.833000000000005e-06,
            "median_process_cpu_seconds": 0.00022,
            "median_process_cpu_wall_ratio": 1.9976985040276178,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 2324631.1010215664,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000107292,
              0.00011175,
              0.000105584,
              0.000113875,
              0.000117459,
              0.000112958,
              0.000108625,
              0.000110125,
              0.000108209,
              0.000106125,
              0.000111458,
              0.000114292,
              0.000106875,
              0.00010925,
              0.000114584
            ],
            "raw_process_cpu_seconds": [
              0.00022,
              0.000218,
              0.000212,
              0.000231,
              0.000236,
              0.000221,
              0.000217,
              0.000217,
              0.000216,
              0.000214,
              0.000227,
              0.000224,
              0.000212,
              0.000226,
              0.000225
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.01707712499774061,
            "speedup_vs_optimized_serial": 1.7688172531214532,
            "speedup_vs_original_serial": 17.928862656072646
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000149834,
            "mad_seconds": 3.543000000000003e-06,
            "median_process_cpu_seconds": 0.000276,
            "median_process_cpu_wall_ratio": 1.8486711856406761,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1708557.4702670956,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000156042,
              0.00015225,
              0.000149834,
              0.000144167,
              0.000148166,
              0.000152542,
              0.000148541,
              0.000177,
              0.000164125,
              0.000149166,
              0.000146291,
              0.00014275,
              0.000169334,
              0.000157875,
              0.0001485
            ],
            "raw_process_cpu_seconds": [
              0.000298,
              0.000286,
              0.000274,
              0.000266,
              0.000272,
              0.000282,
              0.000276,
              0.000329,
              0.000289,
              0.000276,
              0.000274,
              0.000266,
              0.000298,
              0.000288,
              0.000272
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.014791166999202687,
            "speedup_vs_optimized_serial": 1.300045383557804,
            "speedup_vs_original_serial": 13.177356274276868
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 9.2625e-05,
            "mad_seconds": 1.1875000000000004e-05,
            "median_process_cpu_seconds": 0.000326,
            "median_process_cpu_wall_ratio": 3.588516746411483,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 2763832.658569501,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              9.0083e-05,
              0.000120791,
              0.000105,
              7.5333e-05,
              0.0001005,
              9.7791e-05,
              9.1167e-05,
              9.2625e-05,
              0.0001045,
              8.6583e-05,
              0.000115667,
              6.7375e-05,
              0.000105,
              7.0083e-05,
              8.4666e-05
            ],
            "raw_process_cpu_seconds": [
              0.000326,
              0.000375,
              0.000345,
              0.000297,
              0.000359,
              0.000326,
              0.000338,
              0.000309,
              0.000375,
              0.000312,
              0.000356,
              0.000264,
              0.000369,
              0.000271,
              0.000317
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.018005457997787744,
            "speedup_vs_optimized_serial": 2.1030067476383265,
            "speedup_vs_original_serial": 21.316232118758435
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "uniform",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.00016625,
            "mad_seconds": 6.917000000000002e-06,
            "median_process_cpu_seconds": 0.000438,
            "median_process_cpu_wall_ratio": 2.673972602739726,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 1539849.6240601505,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000170208,
              0.00013775,
              0.000128542,
              0.000140042,
              0.000179583,
              0.000164166,
              0.000185,
              0.000159333,
              0.000169,
              0.000168125,
              0.000160958,
              0.00016625,
              0.000179333,
              0.000159667,
              0.0001825
            ],
            "raw_process_cpu_seconds": [
              0.000459,
              0.0004,
              0.000363,
              0.000383,
              0.000434,
              0.000438,
              0.000463,
              0.000419,
              0.000475,
              0.000438,
              0.000471,
              0.000467,
              0.000448,
              0.000423,
              0.000488
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "uniform",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.015119749994482845,
            "speedup_vs_optimized_serial": 1.171675187969925,
            "speedup_vs_original_serial": 11.876186466165414
          },
          {
            "flavor": "optimized",
            "mode": "serial",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 0,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.002048333,
            "mad_seconds": 7.104099999999999e-05,
            "median_process_cpu_seconds": 0.00205,
            "median_process_cpu_wall_ratio": 1.000108912384176,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 124979.67859718124,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.001977292,
              0.001960541,
              0.002154125,
              0.002246208,
              0.002048333,
              0.002033125,
              0.0019885,
              0.002069291,
              0.0022455,
              0.002289084,
              0.002020333,
              0.001909792,
              0.002085208,
              0.002153,
              0.002017458
            ],
            "raw_process_cpu_seconds": [
              0.001982,
              0.001961,
              0.002118,
              0.002215,
              0.00205,
              0.002034,
              0.001989,
              0.002062,
              0.002246,
              0.00229,
              0.002015,
              0.00191,
              0.002078,
              0.002153,
              0.002017
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "serial",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.04942191699956311
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.00203325,
            "mad_seconds": 4.387500000000008e-05,
            "median_process_cpu_seconds": 0.002036,
            "median_process_cpu_wall_ratio": 1.0013525144473134,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 125906.79945899422,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.002002542,
              0.002005959,
              0.001999208,
              0.00203325,
              0.001986958,
              0.002017584,
              0.002074125,
              0.002141291,
              0.001973292,
              0.002188041,
              0.00207325,
              0.002154958,
              0.001983625,
              0.002077125,
              0.002161583
            ],
            "raw_process_cpu_seconds": [
              0.002003,
              0.002012,
              0.002003,
              0.002036,
              0.001993,
              0.002015,
              0.00208,
              0.00213,
              0.001978,
              0.002191,
              0.002076,
              0.00215,
              0.001992,
              0.00208,
              0.002153
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.05712745900382288,
            "speedup_vs_optimized_serial": 1.0074181728759375
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 1,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.002072375,
            "mad_seconds": 3.166699999999998e-05,
            "median_process_cpu_seconds": 0.002075,
            "median_process_cpu_wall_ratio": 1.0016013285095782,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 123529.76657216961,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.002057334,
              0.002070416,
              0.002072375,
              0.002168167,
              0.002122791,
              0.002055334,
              0.002065792,
              0.00210975,
              0.002022417,
              0.002040708,
              0.002123292,
              0.002187875,
              0.002107625,
              0.002096166,
              0.002055625
            ],
            "raw_process_cpu_seconds": [
              0.002069,
              0.002075,
              0.00207,
              0.00215,
              0.002131,
              0.002059,
              0.002064,
              0.002109,
              0.002028,
              0.002054,
              0.002119,
              0.00219,
              0.002111,
              0.002099,
              0.002059
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "1",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.049943375001021195,
            "speedup_vs_optimized_serial": 0.9883988177815308
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.001170584,
            "mad_seconds": 6.191699999999987e-05,
            "median_process_cpu_seconds": 0.002232,
            "median_process_cpu_wall_ratio": 1.9106662361425033,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 218694.2585922924,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.001213958,
              0.001108667,
              0.001170584,
              0.001145917,
              0.00113375,
              0.001156375,
              0.001161375,
              0.001773125,
              0.00130275,
              0.001259958,
              0.001453792,
              0.00124075,
              0.001098292,
              0.001252834,
              0.00113275
            ],
            "raw_process_cpu_seconds": [
              0.002292,
              0.002187,
              0.002242,
              0.00222,
              0.002202,
              0.002227,
              0.002219,
              0.002837,
              0.002297,
              0.002374,
              0.002266,
              0.002272,
              0.002186,
              0.002223,
              0.002232
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "2",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.041597458002797794,
            "speedup_vs_optimized_serial": 1.7498385421293987
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 2,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.001177167,
            "mad_seconds": 2.179200000000008e-05,
            "median_process_cpu_seconds": 0.002277,
            "median_process_cpu_wall_ratio": 1.9372817879225053,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 217471.26788297668,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.001199208,
              0.001159125,
              0.001191667,
              0.001177167,
              0.001273167,
              0.001155375,
              0.001186292,
              0.001162042,
              0.001201792,
              0.001184666,
              0.00111475,
              0.001166583,
              0.001126041,
              0.00114175,
              0.001258125
            ],
            "raw_process_cpu_seconds": [
              0.002306,
              0.002277,
              0.002319,
              0.002281,
              0.002354,
              0.002264,
              0.002277,
              0.002262,
              0.002288,
              0.002283,
              0.002189,
              0.00226,
              0.002194,
              0.002201,
              0.002273
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "2",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.04546154200215824,
            "speedup_vs_optimized_serial": 1.7400530256114892
          },
          {
            "flavor": "optimized",
            "mode": "context",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000580916,
            "mad_seconds": 1.2123999999999967e-05,
            "median_process_cpu_seconds": 0.002219,
            "median_process_cpu_wall_ratio": 3.829681046643682,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 440683.3345957075,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000589834,
              0.000576958,
              0.00063625,
              0.000591625,
              0.000562292,
              0.000580166,
              0.000568417,
              0.000609458,
              0.000582,
              0.000566834,
              0.000568792,
              0.000580916,
              0.000561291,
              0.000589083,
              0.000632208
            ],
            "raw_process_cpu_seconds": [
              0.00225,
              0.002219,
              0.002297,
              0.002208,
              0.002185,
              0.002183,
              0.002209,
              0.002263,
              0.002235,
              0.002176,
              0.00221,
              0.002221,
              0.002182,
              0.002256,
              0.002238
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "context",
              "--workers",
              "4",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.02973924999969313,
            "speedup_vs_optimized_serial": 3.526039909384489
          },
          {
            "flavor": "optimized",
            "mode": "cold",
            "weights": "distance",
            "n": 256,
            "q": 256,
            "d": 1,
            "k": 256,
            "workers_requested": 4,
            "cpu_hard_limit": 12,
            "cpu_auto_workers": 12,
            "seed": 20261030,
            "warmups": 1,
            "median_seconds": 0.000656709,
            "mad_seconds": 1.5415999999999976e-05,
            "median_process_cpu_seconds": 0.002378,
            "median_process_cpu_wall_ratio": 3.5936769558510697,
            "cpu_timing_scope": "process CPU seconds over predict; ratio estimates average CPU equivalents, not instantaneous utilization",
            "rows_per_second": 389822.58504147193,
            "compiler": "Apple LLVM 21.0.0 (clang-2100.3.34.2)",
            "cplusplus": 202002,
            "ndebug": true,
            "fast_math": false,
            "verification": "exact labels,indices,distances; exact uniform probabilities; weighted oracle tolerance2e-14 and exact context/serial probabilities",
            "timing_scope": "predict native call; excludes fit,input creation,verification,process startup; cold includes context creation/destruction",
            "raw_seconds": [
              0.000672125,
              0.000656709,
              0.000652875,
              0.000657708,
              0.000639042,
              0.000690458,
              0.000651625,
              0.000636833,
              0.000678958,
              0.000685542,
              0.000778625,
              0.000673042,
              0.000650167,
              0.000654708,
              0.000646667
            ],
            "raw_process_cpu_seconds": [
              0.002404,
              0.00236,
              0.002397,
              0.002354,
              0.002336,
              0.002442,
              0.002348,
              0.002318,
              0.002359,
              0.002411,
              0.002555,
              0.002412,
              0.002378,
              0.002394,
              0.002339
            ],
            "command": [
              "aner_knn_benchmark",
              "--n",
              "256",
              "--q",
              "256",
              "--d",
              "1",
              "--k",
              "256",
              "--weights",
              "distance",
              "--mode",
              "cold",
              "--workers",
              "4",
              "--repeats",
              "15",
              "--warmups",
              "1",
              "--seed",
              "20261030"
            ],
            "subprocess_wall_seconds": 0.027531916995940264,
            "speedup_vs_optimized_serial": 3.1190877542412236
          }
        ],
        "skipped": []
      }
    }
  ]
}