{
  "evidenceVersion": "local-ollama-speed-v1",
  "capturedAt": "2026-06-02T00:43:40.131Z",
  "hardwareSlug": "rtx-3080-16gb-mobile",
  "modelTag": "gemma3:12b",
  "catalogSlug": "gemma-3-12b",
  "quant": "Q4_K_M",
  "contextSize": 4096,
  "numPredict": 256,
  "prompt": "Write a detailed explanation of how transformer attention works.",
  "scenario": "single-stream",
  "runtime": "ollama-api",
  "command": "POST http://localhost:11434/api/generate model=gemma3:12b temperature=0 top_p=1 seed=42 num_ctx=4096 num_predict=256",
  "env": {
    "os": "Microsoft Windows [Version 10.0.26200.8457]",
    "platform": "win32",
    "arch": "x64",
    "cpu": "AMD Ryzen 9 5900HX with Radeon Graphics        ",
    "cpuThreads": 16,
    "ramGb": 31.9,
    "gpu": {
      "vendor": "nvidia",
      "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
      "driver": "571.96",
      "memoryTotalMb": 16384,
      "memoryUsedMb": 1859,
      "tempC": 53,
      "smClockMhz": 210,
      "powerDrawW": 11.54
    },
    "ollamaVersionCli": "ollama version is 0.24.0",
    "ollamaApiVersion": {
      "version": "0.24.0"
    }
  },
  "ollamaTagDetails": {
    "name": "gemma3:12b",
    "model": "gemma3:12b",
    "modified_at": "2026-06-02T03:42:53.7437177+03:00",
    "size": 8149190253,
    "digest": "f4031aab637d1ffa37b42570452ae0e4fad0314754d17ded67322e4b95836f8a",
    "details": {
      "parent_model": "",
      "format": "gguf",
      "family": "gemma3",
      "families": [
        "gemma3"
      ],
      "parameter_size": "12.2B",
      "quantization_level": "Q4_K_M"
    }
  },
  "notes": null,
  "logHashSha256": "9e3ece081923a4d6d8bb0d2c467565d36f5e18d68ca5fd3ad2879e6662458f4a",
  "runsCaptured": 5,
  "decodeStatsAllRuns": {
    "min": 43.19422363647309,
    "max": 43.60562736753219,
    "median": 43.31408738410711,
    "p5": 43.21762889985217,
    "p95": 43.572791303389565,
    "mean": 43.37332707765999,
    "n": 5
  },
  "decodeStatsSteadyRuns": {
    "min": 43.19422363647309,
    "max": 43.441447046819036,
    "median": 43.31266866873781,
    "p5": 43.2117775840074,
    "p95": 43.422343097412245,
    "mean": 43.31525200519193,
    "n": 4
  },
  "runs": [
    {
      "runIndex": 1,
      "startedAt": "2026-06-02T00:42:58.100Z",
      "finishedAt": "2026-06-02T00:43:11.493Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run1.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run1.log",
      "decodeTokPerSec": 43.60562736753219,
      "prefillTokPerSec": 415.13812082267265,
      "totalMs": 13308.3559,
      "loadMs": 6977.5713,
      "promptEvalMs": 45.7679,
      "evalMs": 5870.8019,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "b9d56a819353590c665a0cc858c1e4bbe39682364f6306dc0e9c6159967a42c8",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 1859,
        "tempC": 53,
        "smClockMhz": 210,
        "powerDrawW": 11.54
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 61,
        "smClockMhz": 1215,
        "powerDrawW": 114.68
      }
    },
    {
      "runIndex": 2,
      "startedAt": "2026-06-02T00:43:11.584Z",
      "finishedAt": "2026-06-02T00:43:18.662Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run2.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run2.log",
      "decodeTokPerSec": 43.441447046819036,
      "prefillTokPerSec": 683.534376382808,
      "totalMs": 6992.1591,
      "loadMs": 765.6987,
      "promptEvalMs": 27.7967,
      "evalMs": 5892.9897,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "01ffb81f99aca86f46568b9da47cdc3de2e441c55bd3fd72ed0d7f90ad9d885f",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 61,
        "smClockMhz": 1215,
        "powerDrawW": 108.29
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 62,
        "smClockMhz": 1230,
        "powerDrawW": 113.39
      }
    },
    {
      "runIndex": 3,
      "startedAt": "2026-06-02T00:43:18.739Z",
      "finishedAt": "2026-06-02T00:43:25.865Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run3.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run3.log",
      "decodeTokPerSec": 43.31408738410711,
      "prefillTokPerSec": 693.754746743004,
      "totalMs": 7039.6447,
      "loadMs": 794.7155,
      "promptEvalMs": 27.3872,
      "evalMs": 5910.3173,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "01ffb81f99aca86f46568b9da47cdc3de2e441c55bd3fd72ed0d7f90ad9d885f",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 61,
        "smClockMhz": 1230,
        "powerDrawW": 106.27
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 64,
        "smClockMhz": 1350,
        "powerDrawW": 113.82
      }
    },
    {
      "runIndex": 4,
      "startedAt": "2026-06-02T00:43:25.942Z",
      "finishedAt": "2026-06-02T00:43:32.991Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run4.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run4.log",
      "decodeTokPerSec": 43.3112499533685,
      "prefillTokPerSec": 691.9759920750539,
      "totalMs": 6949.5291,
      "loadMs": 702.3682,
      "promptEvalMs": 27.4576,
      "evalMs": 5910.7045,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "01ffb81f99aca86f46568b9da47cdc3de2e441c55bd3fd72ed0d7f90ad9d885f",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 64,
        "smClockMhz": 1770,
        "powerDrawW": 106.89
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 66,
        "smClockMhz": 1275,
        "powerDrawW": 114.72
      }
    },
    {
      "runIndex": 5,
      "startedAt": "2026-06-02T00:43:33.069Z",
      "finishedAt": "2026-06-02T00:43:40.130Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run5.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-02/gemma3-12b/run5.log",
      "decodeTokPerSec": 43.19422363647309,
      "prefillTokPerSec": 694.67041544947,
      "totalMs": 6978.741,
      "loadMs": 713.958,
      "promptEvalMs": 27.3511,
      "evalMs": 5926.7184,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "01ffb81f99aca86f46568b9da47cdc3de2e441c55bd3fd72ed0d7f90ad9d885f",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 66,
        "smClockMhz": 1275,
        "powerDrawW": 108.08
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 10750,
        "tempC": 68,
        "smClockMhz": 1260,
        "powerDrawW": 111.74
      }
    }
  ],
  "errors": [],
  "uploadEligible": true
}