Local Inference

← Qwen3.8 27B Q4_K_M · 32k ctx, f16 KV

Run dd42310b

Kindchat benchmark
Started2026-09-17 17:34 UTC
Duration782 s
Enginellama.cpp@91f6a6cf3 (build 10883)
Driver595.99.02
Throttledno
Commandlab bench qwen3.8-27b-q4_k_m-gguf/32k-f16kv --http
Lab version0.1.0+795a2d1

Metrics

MetricMethodPromptGenDepthValue± sdUnit
decode_tpsgreedy—1024030.80.03t/s
decode_tpsdefault sampling—1024030.80.06t/s
e2e_tpsgreedy—1024030.6—t/s
e2e_tpsdefault sampling—1024030.6—t/s
gpu_w_avggreedy—10240250—W
gpu_w_avgdefault sampling—10240249—W
load_schat benchmark——06.20—s
oom_killschat benchmark——00.00—count
out_tokens_meangreedy—10240946—tokens
out_tokens_meandefault sampling—10240948—tokens
prefill_tpsgreedy—10240460—t/s
prefill_tpsdefault sampling—10240459—t/s
ram_peak_mbchat benchmark——024,448—MiB
tokens_per_joulegreedy—102400.12—tok/J
tokens_per_jouledefault sampling—102400.12—tok/J
tpot_msgreedy—1024032.50.04ms
tpot_msdefault sampling—1024032.40.06ms
ttft_msgreedy—1024022040.6ms
ttft_msdefault sampling—1024022041.1ms
vram_peak_mbchat benchmark——018,406—MiB

Raw

{
  "bench": {
    "protocol": "chat-c1-v1",
    "prompts": 8,
    "max_tokens": 1024,
    "concurrency": 1,
    "cohorts": {
      "http-sampled": {
        "temperature": 1,
        "top_p": 0.95,
        "top_k": 20,
        "min_p": 0,
        "presence_penalty": 0
      },
      "http-greedy": {
        "temperature": 0
      }
    },
    "thinking": false,
    "warmup": "all prompts once, discarded"
  },
  "telemetry_summary": {
    "vram_baseline_mb": 0.9375,
    "vram_peak_mb": 18407.125,
    "ram_baseline_mb": 23629.37109375,
    "ram_peak_mb": 24448.46484375,
    "proc_rss_peak_mb": 17065.44140625,
    "temp_max_c": 65,
    "power_max_w": 251.185,
    "clock_max_mhz": 1620,
    "throttled": false,
    "power_capped_frac": 0.9883922615076718,
    "phases": {
      "boot": {
        "duration_s": 6.216221084003337,
        "energy_j": 626.443,
        "power_avg_w": 100.77553412829415
      },
      "warmup": {
        "duration_s": 250.2421405549394,
        "energy_j": 62429.96,
        "power_avg_w": 249.47820483614277
      },
      "http-sampled": {
        "duration_s": 247.43428473907989,
        "energy_j": 61730.746,
        "power_avg_w": 249.48339744063858
      },
      "http-greedy": {
        "duration_s": 247.1656156849349,
        "energy_j": 61685.715,
        "power_avg_w": 249.57239634266747
      }
    }
  },
  "protocol": "chat-c1-v1",
  "requests": [
    {
      "phase": "warmup",
      "prompt": 0,
      "prompt_tokens": 60,
      "completion_tokens": 1024,
      "ttft_ms": 284.02,
      "tpot_ms": 31.769,
      "decode_tps": 31.48,
      "e2e_s": 32.784,
      "finish_reason": "length",
      "server_decode_tps": 31.48
    },
    {
      "phase": "warmup",
      "prompt": 1,
      "prompt_tokens": 49,
      "completion_tokens": 1024,
      "ttft_ms": 190.69,
      "tpot_ms": 32.451,
      "decode_tps": 30.82,
      "e2e_s": 33.389,
      "finish_reason": "length",
      "server_decode_tps": 30.82
    },
    {
      "phase": "warmup",
      "prompt": 2,
      "prompt_tokens": 68,
      "completion_tokens": 1024,
      "ttft_ms": 196.78,
      "tpot_ms": 33.171,
      "decode_tps": 30.15,
      "e2e_s": 34.131,
      "finish_reason": "length",
      "server_decode_tps": 30.15
    },
    {
      "phase": "warmup",
      "prompt": 3,
      "prompt_tokens": 196,
      "completion_tokens": 1024,
      "ttft_ms": 268.26,
      "tpot_ms": 33.294,
      "decode_tps": 30.04,
      "e2e_s": 34.328,
      "finish_reason": "length",
      "server_decode_tps": 30.04
    },
    {
      "phase": "warmup",
      "prompt": 4,
      "prompt_tokens": 92,
      "completion_tokens": 1024,
      "ttft_ms": 212.95,
      "tpot_ms": 33.01,
      "decode_tps": 30.29,
      "e2e_s": 33.982,
      "finish_reason": "length",
      "server_decode_tps": 30.29
    },
    {
      "phase": "warmup",
      "prompt": 5,
      "prompt_tokens": 55,
      "completion_tokens": 1024,
      "ttft_ms": 195.77,
      "tpot_ms": 32.829,
      "decode_tps": 30.46,
      "e2e_s": 33.781,
      "finish_reason": "length",
      "server_decode_tps": 30.46
    },
    {
      "phase": "warmup",
      "prompt": 6,
      "prompt_tokens": 50,
      "completion_tokens": 1024,
      "ttft_ms": 191.74,
      "tpot_ms": 32.659,
      "decode_tps": 30.62,
      "e2e_s": 33.602,
      "finish_reason": "length",
      "server_decode_tps": 30.62
    },
    {
      "phase": "warmup",
      "prompt": 7,
      "prompt_tokens": 240,
      "completion_tokens": 429,
      "ttft_ms": 305.6,
      "tpot_ms": 32.484,
      "decode_tps": 30.78,
      "e2e_s": 14.242,
      "finish_reason": "stop",
      "server_decode_tps": 30.71
    },
    {
      "phase": "http-sampled",
      "prompt": 0,
      "prompt_tokens": 60,
      "completion_tokens": 1024,
      "ttft_ms": 195.28,
      "tpot_ms": 32.51,
      "decode_tps": 30.76,
      "e2e_s": 33.453,
      "finish_reason": "length",
      "server_decode_tps": 30.76
    },
    {
      "phase": "http-sampled",
      "prompt": 1,
      "prompt_tokens": 49,
      "completion_tokens": 1024,
      "ttft_ms": 192.46,
      "tpot_ms": 32.369,
      "decode_tps": 30.89,
      "e2e_s": 33.306,
      "finish_reason": "length",
      "server_decode_tps": 30.89
    },
    {
      "phase": "http-sampled",
      "prompt": 2,
      "prompt_tokens": 68,
      "completion_tokens": 1024,
      "ttft_ms": 198.46,
      "tpot_ms": 32.345,
      "decode_tps": 30.92,
      "e2e_s": 33.288,
      "finish_reason": "length",
      "server_decode_tps": 30.92
    },
    {
      "phase": "http-sampled",
      "prompt": 3,
      "prompt_tokens": 196,
      "completion_tokens": 1024,
      "ttft_ms": 264.01,
      "tpot_ms": 32.454,
      "decode_tps": 30.81,
      "e2e_s": 33.465,
      "finish_reason": "length",
      "server_decode_tps": 30.81
    },
    {
      "phase": "http-sampled",
      "prompt": 4,
      "prompt_tokens": 92,
      "completion_tokens": 1024,
      "ttft_ms": 215.9,
      "tpot_ms": 32.481,
      "decode_tps": 30.79,
      "e2e_s": 33.444,
      "finish_reason": "length",
      "server_decode_tps": 30.79
    },
    {
      "phase": "http-sampled",
      "prompt": 5,
      "prompt_tokens": 55,
      "completion_tokens": 1024,
      "ttft_ms": 200.97,
      "tpot_ms": 32.469,
      "decode_tps": 30.8,
      "e2e_s": 33.417,
      "finish_reason": "length",
      "server_decode_tps": 30.8
    },
    {
      "phase": "http-sampled",
      "prompt": 6,
      "prompt_tokens": 50,
      "completion_tokens": 1024,
      "ttft_ms": 193.15,
      "tpot_ms": 32.478,
      "decode_tps": 30.79,
      "e2e_s": 33.418,
      "finish_reason": "length",
      "server_decode_tps": 30.79
    },
    {
      "phase": "http-sampled",
      "prompt": 7,
      "prompt_tokens": 240,
      "completion_tokens": 412,
      "ttft_ms": 303.47,
      "tpot_ms": 32.361,
      "decode_tps": 30.9,
      "e2e_s": 13.637,
      "finish_reason": "stop",
      "server_decode_tps": 30.83
    },
    {
      "phase": "http-greedy",
      "prompt": 0,
      "prompt_tokens": 60,
      "completion_tokens": 1024,
      "ttft_ms": 195.67,
      "tpot_ms": 32.453,
      "decode_tps": 30.81,
      "e2e_s": 33.396,
      "finish_reason": "length",
      "server_decode_tps": 30.81
    },
    {
      "phase": "http-greedy",
      "prompt": 1,
      "prompt_tokens": 49,
      "completion_tokens": 1024,
      "ttft_ms": 191.68,
      "tpot_ms": 32.463,
      "decode_tps": 30.8,
      "e2e_s": 33.402,
      "finish_reason": "length",
      "server_decode_tps": 30.8
    },
    {
      "phase": "http-greedy",
      "prompt": 2,
      "prompt_tokens": 68,
      "completion_tokens": 1024,
      "ttft_ms": 198.58,
      "tpot_ms": 32.442,
      "decode_tps": 30.82,
      "e2e_s": 33.387,
      "finish_reason": "length",
      "server_decode_tps": 30.82
    },
    {
      "phase": "http-greedy",
      "prompt": 3,
      "prompt_tokens": 196,
      "completion_tokens": 1024,
      "ttft_ms": 262.74,
      "tpot_ms": 32.462,
      "decode_tps": 30.81,
      "e2e_s": 33.472,
      "finish_reason": "length",
      "server_decode_tps": 30.8
    },
    {
      "phase": "http-greedy",
      "prompt": 4,
      "prompt_tokens": 92,
      "completion_tokens": 1024,
      "ttft_ms": 214.53,
      "tpot_ms": 32.48,
      "decode_tps": 30.79,
      "e2e_s": 33.442,
      "finish_reason": "length",
      "server_decode_tps": 30.79
    },
    {
      "phase": "http-greedy",
      "prompt": 5,
      "prompt_tokens": 55,
      "completion_tokens": 1024,
      "ttft_ms": 200.12,
      "tpot_ms": 32.523,
      "decode_tps": 30.75,
      "e2e_s": 33.472,
      "finish_reason": "length",
      "server_decode_tps": 30.75
    },
    {
      "phase": "http-greedy",
      "prompt": 6,
      "prompt_tokens": 50,
      "completion_tokens": 1024,
      "ttft_ms": 195.24,
      "tpot_ms": 32.51,
      "decode_tps": 30.76,
      "e2e_s": 33.454,
      "finish_reason": "length",
      "server_decode_tps": 30.76
    },
    {
      "phase": "http-greedy",
      "prompt": 7,
      "prompt_tokens": 240,
      "completion_tokens": 396,
      "ttft_ms": 302.48,
      "tpot_ms": 32.408,
      "decode_tps": 30.86,
      "e2e_s": 13.136,
      "finish_reason": "stop",
      "server_decode_tps": 30.78
    }
  ],
  "server": {
    "cmdline": [
      "llama-server",
      "-m",
      "Qwen3.8-27B-Q4_K_M.gguf",
      "--host",
      "127.0.0.1",
      "--port",
      "8081",
      "-a",
      "qwen3.8-27b-q4_k_m-gguf/32k-f16kv",
      "--fit",
      "off",
      "-c",
      "32768",
      "-np",
      "1",
      "--cache-ram",
      "0",
      "-ngl",
      "999",
      "-fa",
      "on",
      "-ctk",
      "f16",
      "-ctv",
      "f16",
      "-b",
      "2048",
      "-ub",
      "512",
      "-t",
      "6",
      "-ncmoe",
      "0",
      "-lm",
      "mmap",
      "--metrics",
      "--temp",
      "1.0",
      "--top-p",
      "0.95",
      "--top-k",
      "20",
      "--min-p",
      "0",
      "--presence-penalty",
      "0"
    ],
    "env": {}
  }
}