{
  "evidenceVersion": "local-ollama-speed-v1",
  "capturedAt": "2026-06-01T21:40:05.927Z",
  "hardwareSlug": "rtx-3080-16gb-mobile",
  "modelTag": "phi3.5:3.8b",
  "catalogSlug": "phi-3.5-mini-instruct",
  "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=phi3.5:3.8b 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": 65,
      "smClockMhz": 210,
      "powerDrawW": 18.64
    },
    "ollamaVersionCli": "ollama version is 0.24.0",
    "ollamaApiVersion": {
      "version": "0.24.0"
    }
  },
  "ollamaTagDetails": {
    "name": "phi3.5:3.8b",
    "model": "phi3.5:3.8b",
    "modified_at": "2026-06-02T00:39:51.0608612+03:00",
    "size": 2176178843,
    "digest": "61819fb370a3c1a9be6694869331e5f85f867a079e9271d66cb223acb81d04ba",
    "details": {
      "parent_model": "",
      "format": "gguf",
      "family": "phi3",
      "families": [
        "phi3"
      ],
      "parameter_size": "3.8B",
      "quantization_level": "Q4_0"
    }
  },
  "notes": null,
  "logHashSha256": "8f494f4aa9db60d8ec59160449e1a872cebad444d6ad4749b220cecfc1d040b0",
  "runsCaptured": 5,
  "decodeStatsAllRuns": {
    "min": 152.18864210219124,
    "max": 157.3729149164565,
    "median": 154.69995068334777,
    "p5": 152.4838264223701,
    "p95": 157.34119775878023,
    "mean": 155.02808010663122,
    "n": 5
  },
  "decodeStatsSteadyRuns": {
    "min": 152.18864210219124,
    "max": 157.3729149164565,
    "median": 155.43944641558033,
    "p5": 152.41003034232537,
    "p95": 157.3491270481993,
    "mean": 155.1101124624521,
    "n": 4
  },
  "runs": [
    {
      "runIndex": 1,
      "startedAt": "2026-06-01T21:39:53.177Z",
      "finishedAt": "2026-06-01T21:39:57.007Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run1.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run1.log",
      "decodeTokPerSec": 154.69995068334777,
      "prefillTokPerSec": 343.0672960807992,
      "totalMs": 3745.1007,
      "loadMs": 1699.8923,
      "promptEvalMs": 58.2976,
      "evalMs": 1654.8163,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "c43519b8e4bc425081e9819ea8e5f4b4b33b2d90077391501eb8c8af299fc1fa",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 4847,
        "tempC": 65,
        "smClockMhz": 210,
        "powerDrawW": 18.66
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 71,
        "smClockMhz": 1395,
        "powerDrawW": 110.88
      }
    },
    {
      "runIndex": 2,
      "startedAt": "2026-06-01T21:39:57.082Z",
      "finishedAt": "2026-06-01T21:39:59.237Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run2.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run2.log",
      "decodeTokPerSec": 153.6645637030855,
      "prefillTokPerSec": 1875.7327080890973,
      "totalMs": 2057.5492,
      "loadMs": 54.0116,
      "promptEvalMs": 10.6625,
      "evalMs": 1665.9664,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "a13aa703a3ab394c847b7392af81dcf342f9ba85d59bd66435ee642119de0a35",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 71,
        "smClockMhz": 1395,
        "powerDrawW": 110.88
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 72,
        "smClockMhz": 1380,
        "powerDrawW": 112.21
      }
    },
    {
      "runIndex": 3,
      "startedAt": "2026-06-01T21:39:59.317Z",
      "finishedAt": "2026-06-01T21:40:01.478Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run3.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run3.log",
      "decodeTokPerSec": 152.18864210219124,
      "prefillTokPerSec": 2164.033758926639,
      "totalMs": 2065.5918,
      "loadMs": 57.3533,
      "promptEvalMs": 9.242,
      "evalMs": 1682.1229,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "a13aa703a3ab394c847b7392af81dcf342f9ba85d59bd66435ee642119de0a35",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 72,
        "smClockMhz": 1380,
        "powerDrawW": 106.24
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 72,
        "smClockMhz": 1410,
        "powerDrawW": 114.27
      }
    },
    {
      "runIndex": 4,
      "startedAt": "2026-06-01T21:40:01.553Z",
      "finishedAt": "2026-06-01T21:40:03.710Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run4.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run4.log",
      "decodeTokPerSec": 157.3729149164565,
      "prefillTokPerSec": 2147.0285125386467,
      "totalMs": 2068.081,
      "loadMs": 56.965,
      "promptEvalMs": 9.3152,
      "evalMs": 1626.7094,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "a13aa703a3ab394c847b7392af81dcf342f9ba85d59bd66435ee642119de0a35",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 72,
        "smClockMhz": 1410,
        "powerDrawW": 108.79
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 73,
        "smClockMhz": 1320,
        "powerDrawW": 111.14
      }
    },
    {
      "runIndex": 5,
      "startedAt": "2026-06-01T21:40:03.795Z",
      "finishedAt": "2026-06-01T21:40:05.926Z",
      "rawPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run5.ollama.json",
      "logPath": "public/benchmarks/evidence/local-ollama-rtx-3080-16gb-mobile-2026-06-01/phi3-5-3-8b/run5.log",
      "decodeTokPerSec": 157.2143291280752,
      "prefillTokPerSec": 2036.18297140181,
      "totalMs": 2049.5266,
      "loadMs": 58.676,
      "promptEvalMs": 9.8223,
      "evalMs": 1628.3503,
      "promptTokens": 20,
      "responseTokens": 256,
      "responseHashSha256": "a13aa703a3ab394c847b7392af81dcf342f9ba85d59bd66435ee642119de0a35",
      "beforeGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 73,
        "smClockMhz": 1320,
        "powerDrawW": 105.22
      },
      "afterGpu": {
        "vendor": "nvidia",
        "name": "NVIDIA GeForce RTX 3080 Laptop GPU",
        "driver": "571.96",
        "memoryTotalMb": 16384,
        "memoryUsedMb": 8720,
        "tempC": 74,
        "smClockMhz": 1350,
        "powerDrawW": 112.3
      }
    }
  ],
  "errors": [],
  "uploadEligible": true
}