{
  "evidenceVersion": "local-ollama-speed-v1",
  "capturedAt": "2026-06-01T21:41:59.093Z",
  "hardwareSlug": "rtx-3080-16gb-mobile",
  "modelTag": "gemma3:4b",
  "catalogSlug": "gemma-3-4b",
  "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=gemma3:4b 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": 4847,
      "tempC": 63,
      "smClockMhz": 210,
      "powerDrawW": 18.1
    },
    "ollamaVersionCli": "ollama version is 0.24.0",
    "ollamaApiVersion": {
      "version": "0.24.0"
    }
  },
  "ollamaTagDetails": {
    "name": "gemma3:4b",
    "model": "gemma3:4b",
    "modified_at": "2026-06-02T00:41:32.2988853+03:00",
    "size": 3338801804,
    "digest": "a2af6cc3eb7fa8be8504abaf9b04e88f17a119ec3f04a3addf55f92841195f5a",
    "details": {
      "parent_model": "",
      "format": "gguf",
      "family": "gemma3",
      "families": [
        "gemma3"
      ],
      "parameter_size": "4.3B",
      "quantization_level": "Q4_K_M"
    }
  },
  "notes": null,
  "logHashSha256": "093986892faf5744f43190370c951328c1f98edac93358ab5f3bfcb7987ba171",
  "runsCaptured": 5,
  "decodeStatsAllRuns": {
    "min": 96.43889582587595,
    "max": 98.29404163405398,
    "median": 97.5597895742045,
    "p5": 96.5716423128824,
    "p95": 98.18507677867224,
    "mean": 97.4289145304376,
    "n": 5
  },
  "decodeStatsSteadyRuns": {
    "min": 97.10262826090823,
    "max": 98.29404163405398,
    "median": 97.65450346567488,
    "p5": 97.17120245790268,
    "p95": 98.21231799251767,
    "mean": 97.676419206578,
    "n": 4
  },
  "runs": [
    {
      "runIndex": 1,
      "startedAt": "2026-06-01T21:41:36.433Z",
      "finishedAt": "2026-06-01T21:41:44.043Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run1.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run1.log",
      "decodeTokPerSec": 96.43889582587595,
      "prefillTokPerSec": 734.1973677092269,
      "totalMs": 7525.9138,
      "loadMs": 4562.8138,
      "promptEvalMs": 25.8786,
      "evalMs": 2654.5306,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "a26d8da45d6427187ea78abbe8599beb9b39b9905c26069995583985409ae23f",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 4847,
        "tempC": 63,
        "smClockMhz": 210,
        "powerDrawW": 18.11
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 70,
        "smClockMhz": 1500,
        "powerDrawW": 114.34
      }
    },
    {
      "runIndex": 2,
      "startedAt": "2026-06-01T21:41:44.120Z",
      "finishedAt": "2026-06-01T21:41:47.789Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run2.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run2.log",
      "decodeTokPerSec": 97.74921735714526,
      "prefillTokPerSec": 1396.1041346726136,
      "totalMs": 3573.4871,
      "loadMs": 715.6797,
      "promptEvalMs": 13.6093,
      "evalMs": 2618.9468,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "d5e0634e53eabcb893b486cb27275d5a9193a2da1e25b9ce70cff98352bdbfa0",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 68,
        "smClockMhz": 1500,
        "powerDrawW": 110.14
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 71,
        "smClockMhz": 1440,
        "powerDrawW": 112.66
      }
    },
    {
      "runIndex": 3,
      "startedAt": "2026-06-01T21:41:47.868Z",
      "finishedAt": "2026-06-01T21:41:51.551Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run3.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run3.log",
      "decodeTokPerSec": 98.29404163405398,
      "prefillTokPerSec": 1340.8420488066506,
      "totalMs": 3616.5934,
      "loadMs": 740.9103,
      "promptEvalMs": 14.1702,
      "evalMs": 2604.4305,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "d5e0634e53eabcb893b486cb27275d5a9193a2da1e25b9ce70cff98352bdbfa0",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 71,
        "smClockMhz": 1755,
        "powerDrawW": 106.54
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 72,
        "smClockMhz": 1515,
        "powerDrawW": 111.07
      }
    },
    {
      "runIndex": 4,
      "startedAt": "2026-06-01T21:41:51.635Z",
      "finishedAt": "2026-06-01T21:41:55.323Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run4.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run4.log",
      "decodeTokPerSec": 97.10262826090823,
      "prefillTokPerSec": 1404.016966436605,
      "totalMs": 3600.3794,
      "loadMs": 733.5162,
      "promptEvalMs": 13.5326,
      "evalMs": 2636.3859,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "d5e0634e53eabcb893b486cb27275d5a9193a2da1e25b9ce70cff98352bdbfa0",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 70,
        "smClockMhz": 1515,
        "powerDrawW": 111.07
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 72,
        "smClockMhz": 1710,
        "powerDrawW": 114.18
      }
    },
    {
      "runIndex": 5,
      "startedAt": "2026-06-01T21:41:55.419Z",
      "finishedAt": "2026-06-01T21:41:59.091Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run5.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/gemma3-4b/run5.log",
      "decodeTokPerSec": 97.5597895742045,
      "prefillTokPerSec": 1246.9728094297398,
      "totalMs": 3589.5427,
      "loadMs": 724.6412,
      "promptEvalMs": 15.2369,
      "evalMs": 2624.0319,
      "promptTokens": 19,
      "responseTokens": 256,
      "responseHashSha256": "d5e0634e53eabcb893b486cb27275d5a9193a2da1e25b9ce70cff98352bdbfa0",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 72,
        "smClockMhz": 1710,
        "powerDrawW": 108.88
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 72,
        "smClockMhz": 1485,
        "powerDrawW": 113.46
      }
    }
  ],
  "errors": [],
  "uploadEligible": true
}