{
  "evidenceVersion": "local-ollama-speed-v1",
  "capturedAt": "2026-06-01T21:48:01.932Z",
  "hardwareSlug": "rtx-3080-16gb-mobile",
  "modelTag": "gemma4:e4b",
  "catalogSlug": "gemma-4-e4b",
  "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=gemma4:e4b 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": 1873,
      "tempC": 60,
      "smClockMhz": 210,
      "powerDrawW": 17.45
    },
    "ollamaVersionCli": "ollama version is 0.24.0",
    "ollamaApiVersion": {
      "version": "0.24.0"
    }
  },
  "ollamaTagDetails": {
    "name": "gemma4:e4b",
    "model": "gemma4:e4b",
    "modified_at": "2026-06-02T00:47:26.1305802+03:00",
    "size": 9608350718,
    "digest": "c6eb396dbd5992bbe3f5cdb947e8bbc0ee413d7c17e2beaae69f5d569cf982eb",
    "details": {
      "parent_model": "",
      "format": "gguf",
      "family": "gemma4",
      "families": [
        "gemma4"
      ],
      "parameter_size": "8.0B",
      "quantization_level": "Q4_K_M"
    }
  },
  "notes": null,
  "logHashSha256": "4e287dd0c35a53c49196f16e0587dc3fde424522104bbb613e9528ababec0fcf",
  "runsCaptured": 5,
  "decodeStatsAllRuns": {
    "min": 77.8974077750985,
    "max": 79.31995286287864,
    "median": 78.18087518053751,
    "p5": 77.92650501203211,
    "p95": 79.14073223470561,
    "mean": 78.37299590005894,
    "n": 5
  },
  "decodeStatsSteadyRuns": {
    "min": 77.8974077750985,
    "max": 78.4238497220135,
    "median": 78.11188457015204,
    "p5": 77.91923070279871,
    "p95": 78.3874035407921,
    "mean": 78.13625665935402,
    "n": 4
  },
  "runs": [
    {
      "runIndex": 1,
      "startedAt": "2026-06-01T21:47:30.302Z",
      "finishedAt": "2026-06-01T21:47:43.914Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run1.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run1.log",
      "decodeTokPerSec": 79.31995286287864,
      "prefillTokPerSec": 201.06922426719936,
      "totalMs": 13532.273,
      "loadMs": 9528.6848,
      "promptEvalMs": 129.3087,
      "evalMs": 3227.4351,
      "promptTokens": 26,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 1873,
        "tempC": 60,
        "smClockMhz": 210,
        "powerDrawW": 17.48
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 67,
        "smClockMhz": 1545,
        "powerDrawW": 114.6
      }
    },
    {
      "runIndex": 2,
      "startedAt": "2026-06-01T21:47:44.005Z",
      "finishedAt": "2026-06-01T21:47:48.424Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run2.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run2.log",
      "decodeTokPerSec": 78.18087518053751,
      "prefillTokPerSec": 1434.6252317471528,
      "totalMs": 4323.175,
      "loadMs": 775.3694,
      "promptEvalMs": 18.1232,
      "evalMs": 3274.4581,
      "promptTokens": 26,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 67,
        "smClockMhz": 1545,
        "powerDrawW": 108.42
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 68,
        "smClockMhz": 1530,
        "powerDrawW": 114.57
      }
    },
    {
      "runIndex": 3,
      "startedAt": "2026-06-01T21:47:48.506Z",
      "finishedAt": "2026-06-01T21:47:52.914Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run3.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run3.log",
      "decodeTokPerSec": 78.4238497220135,
      "prefillTokPerSec": 1441.8890965455664,
      "totalMs": 4314.8662,
      "loadMs": 769.4299,
      "promptEvalMs": 18.0319,
      "evalMs": 3264.3131,
      "promptTokens": 26,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 68,
        "smClockMhz": 1530,
        "powerDrawW": 108.5
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 69,
        "smClockMhz": 1545,
        "powerDrawW": 114.69
      }
    },
    {
      "runIndex": 4,
      "startedAt": "2026-06-01T21:47:52.989Z",
      "finishedAt": "2026-06-01T21:47:57.332Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run4.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run4.log",
      "decodeTokPerSec": 78.04289395976657,
      "prefillTokPerSec": 1439.542004174672,
      "totalMs": 4257.7028,
      "loadMs": 707.0714,
      "promptEvalMs": 18.0613,
      "evalMs": 3280.2474,
      "promptTokens": 26,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 69,
        "smClockMhz": 1545,
        "powerDrawW": 108.14
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 70,
        "smClockMhz": 1530,
        "powerDrawW": 114.7
      }
    },
    {
      "runIndex": 5,
      "startedAt": "2026-06-01T21:47:57.411Z",
      "finishedAt": "2026-06-01T21:48:01.929Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run5.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma4-e4b/run5.log",
      "decodeTokPerSec": 77.8974077750985,
      "prefillTokPerSec": 1472.1786545419543,
      "totalMs": 4424.1151,
      "loadMs": 856.3666,
      "promptEvalMs": 17.6609,
      "evalMs": 3286.3738,
      "promptTokens": 26,
      "responseTokens": 256,
      "responseHashSha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 70,
        "smClockMhz": 1530,
        "powerDrawW": 109.96
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 11576,
        "tempC": 71,
        "smClockMhz": 1530,
        "powerDrawW": 114.53
      }
    }
  ],
  "errors": [],
  "uploadEligible": true
}