Local Inference

← Qwen3.8 27B Q4_K_M · 64k ctx, q8_0 KV

Run e75bd3f1

Kindchat benchmark
Started2026-09-17 17:21 UTC
Duration766 s
Enginellama.cpp@91f6a6cf3 (build 10883)
Driver595.99.02
Throttledno
Commandlab bench qwen3.8-27b-q4_k_m-gguf/64k-q8kv --http
Lab version0.1.0+795a2d1

Metrics

MetricMethodPromptGenDepthValue± sdUnit
decode_tpsgreedy—1024030.50.03t/s
decode_tpsdefault sampling—1024030.60.05t/s
e2e_tpsgreedy—1024030.3—t/s
e2e_tpsdefault sampling—1024030.4—t/s
gpu_w_avggreedy—10240250—W
gpu_w_avgdefault sampling—10240249—W
load_schat benchmark——018.1—s
oom_killschat benchmark——00.00—count
out_tokens_meangreedy—10240948—tokens
out_tokens_meandefault sampling—10240947—tokens
prefill_tpsgreedy—10240449—t/s
prefill_tpsdefault sampling—10240450—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—1024032.80.04ms
tpot_msdefault sampling—1024032.70.05ms
ttft_msgreedy—1024022540.9ms
ttft_msdefault sampling—1024022539.9ms
vram_peak_mbchat benchmark——018,780—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": 18781.125,
    "ram_baseline_mb": 24125.5390625,
    "ram_peak_mb": 24448.44921875,
    "proc_rss_peak_mb": 15609.2265625,
    "temp_max_c": 65,
    "power_max_w": 252.307,
    "clock_max_mhz": 1620,
    "throttled": false,
    "power_capped_frac": 0.9732283464566929,
    "phases": {
      "boot": {
        "duration_s": 18.103008003905416,
        "energy_j": 1840.731,
        "power_avg_w": 101.68094714441344
      },
      "warmup": {
        "duration_s": 246.06503996800166,
        "energy_j": 61381.046,
        "power_avg_w": 249.45049490972795
      },
      "http-sampled": {
        "duration_s": 249.32226216094568,
        "energy_j": 62198.215,
        "power_avg_w": 249.4691587542592
      },
      "http-greedy": {
        "duration_s": 250.10854888509493,
        "energy_j": 62404.232,
        "power_avg_w": 249.50859248185796
      }
    }
  },
  "protocol": "chat-c1-v1",
  "requests": [
    {
      "phase": "warmup",
      "prompt": 0,
      "prompt_tokens": 60,
      "completion_tokens": 1024,
      "ttft_ms": 330.72,
      "tpot_ms": 31.386,
      "decode_tps": 31.86,
      "e2e_s": 32.441,
      "finish_reason": "length",
      "server_decode_tps": 31.86
    },
    {
      "phase": "warmup",
      "prompt": 1,
      "prompt_tokens": 49,
      "completion_tokens": 1024,
      "ttft_ms": 193.81,
      "tpot_ms": 32.16,
      "decode_tps": 31.09,
      "e2e_s": 33.094,
      "finish_reason": "length",
      "server_decode_tps": 31.09
    },
    {
      "phase": "warmup",
      "prompt": 2,
      "prompt_tokens": 68,
      "completion_tokens": 1024,
      "ttft_ms": 200.86,
      "tpot_ms": 32.963,
      "decode_tps": 30.34,
      "e2e_s": 33.922,
      "finish_reason": "length",
      "server_decode_tps": 30.34
    },
    {
      "phase": "warmup",
      "prompt": 3,
      "prompt_tokens": 196,
      "completion_tokens": 921,
      "ttft_ms": 270.76,
      "tpot_ms": 33.453,
      "decode_tps": 29.89,
      "e2e_s": 31.081,
      "finish_reason": "stop",
      "server_decode_tps": 29.86
    },
    {
      "phase": "warmup",
      "prompt": 4,
      "prompt_tokens": 92,
      "completion_tokens": 1024,
      "ttft_ms": 219.51,
      "tpot_ms": 33.533,
      "decode_tps": 29.82,
      "e2e_s": 34.524,
      "finish_reason": "length",
      "server_decode_tps": 29.82
    },
    {
      "phase": "warmup",
      "prompt": 5,
      "prompt_tokens": 55,
      "completion_tokens": 1024,
      "ttft_ms": 199.2,
      "tpot_ms": 33.177,
      "decode_tps": 30.14,
      "e2e_s": 34.139,
      "finish_reason": "length",
      "server_decode_tps": 30.14
    },
    {
      "phase": "warmup",
      "prompt": 6,
      "prompt_tokens": 50,
      "completion_tokens": 1024,
      "ttft_ms": 195.99,
      "tpot_ms": 32.979,
      "decode_tps": 30.32,
      "e2e_s": 33.934,
      "finish_reason": "length",
      "server_decode_tps": 30.32
    },
    {
      "phase": "warmup",
      "prompt": 7,
      "prompt_tokens": 240,
      "completion_tokens": 385,
      "ttft_ms": 304.72,
      "tpot_ms": 32.782,
      "decode_tps": 30.5,
      "e2e_s": 12.926,
      "finish_reason": "stop",
      "server_decode_tps": 30.43
    },
    {
      "phase": "http-sampled",
      "prompt": 0,
      "prompt_tokens": 60,
      "completion_tokens": 1024,
      "ttft_ms": 202.22,
      "tpot_ms": 32.731,
      "decode_tps": 30.55,
      "e2e_s": 33.686,
      "finish_reason": "length",
      "server_decode_tps": 30.55
    },
    {
      "phase": "http-sampled",
      "prompt": 1,
      "prompt_tokens": 49,
      "completion_tokens": 1024,
      "ttft_ms": 199.1,
      "tpot_ms": 32.657,
      "decode_tps": 30.62,
      "e2e_s": 33.607,
      "finish_reason": "length",
      "server_decode_tps": 30.62
    },
    {
      "phase": "http-sampled",
      "prompt": 2,
      "prompt_tokens": 68,
      "completion_tokens": 1024,
      "ttft_ms": 204.16,
      "tpot_ms": 32.601,
      "decode_tps": 30.67,
      "e2e_s": 33.555,
      "finish_reason": "length",
      "server_decode_tps": 30.67
    },
    {
      "phase": "http-sampled",
      "prompt": 3,
      "prompt_tokens": 196,
      "completion_tokens": 1024,
      "ttft_ms": 270.74,
      "tpot_ms": 32.701,
      "decode_tps": 30.58,
      "e2e_s": 33.724,
      "finish_reason": "length",
      "server_decode_tps": 30.58
    },
    {
      "phase": "http-sampled",
      "prompt": 4,
      "prompt_tokens": 92,
      "completion_tokens": 1024,
      "ttft_ms": 220.9,
      "tpot_ms": 32.723,
      "decode_tps": 30.56,
      "e2e_s": 33.697,
      "finish_reason": "length",
      "server_decode_tps": 30.56
    },
    {
      "phase": "http-sampled",
      "prompt": 5,
      "prompt_tokens": 55,
      "completion_tokens": 1024,
      "ttft_ms": 201.84,
      "tpot_ms": 32.691,
      "decode_tps": 30.59,
      "e2e_s": 33.645,
      "finish_reason": "length",
      "server_decode_tps": 30.59
    },
    {
      "phase": "http-sampled",
      "prompt": 6,
      "prompt_tokens": 50,
      "completion_tokens": 1024,
      "ttft_ms": 198.5,
      "tpot_ms": 32.735,
      "decode_tps": 30.55,
      "e2e_s": 33.687,
      "finish_reason": "length",
      "server_decode_tps": 30.55
    },
    {
      "phase": "http-sampled",
      "prompt": 7,
      "prompt_tokens": 240,
      "completion_tokens": 411,
      "ttft_ms": 303.68,
      "tpot_ms": 32.634,
      "decode_tps": 30.64,
      "e2e_s": 13.717,
      "finish_reason": "stop",
      "server_decode_tps": 30.57
    },
    {
      "phase": "http-greedy",
      "prompt": 0,
      "prompt_tokens": 60,
      "completion_tokens": 1024,
      "ttft_ms": 201.39,
      "tpot_ms": 32.791,
      "decode_tps": 30.5,
      "e2e_s": 33.747,
      "finish_reason": "length",
      "server_decode_tps": 30.5
    },
    {
      "phase": "http-greedy",
      "prompt": 1,
      "prompt_tokens": 49,
      "completion_tokens": 1024,
      "ttft_ms": 195.6,
      "tpot_ms": 32.785,
      "decode_tps": 30.5,
      "e2e_s": 33.735,
      "finish_reason": "length",
      "server_decode_tps": 30.5
    },
    {
      "phase": "http-greedy",
      "prompt": 2,
      "prompt_tokens": 68,
      "completion_tokens": 1024,
      "ttft_ms": 204.84,
      "tpot_ms": 32.734,
      "decode_tps": 30.55,
      "e2e_s": 33.692,
      "finish_reason": "length",
      "server_decode_tps": 30.55
    },
    {
      "phase": "http-greedy",
      "prompt": 3,
      "prompt_tokens": 196,
      "completion_tokens": 1024,
      "ttft_ms": 270.66,
      "tpot_ms": 32.775,
      "decode_tps": 30.51,
      "e2e_s": 33.8,
      "finish_reason": "length",
      "server_decode_tps": 30.51
    },
    {
      "phase": "http-greedy",
      "prompt": 4,
      "prompt_tokens": 92,
      "completion_tokens": 1024,
      "ttft_ms": 218.66,
      "tpot_ms": 32.743,
      "decode_tps": 30.54,
      "e2e_s": 33.715,
      "finish_reason": "length",
      "server_decode_tps": 30.54
    },
    {
      "phase": "http-greedy",
      "prompt": 5,
      "prompt_tokens": 55,
      "completion_tokens": 1024,
      "ttft_ms": 204.7,
      "tpot_ms": 32.769,
      "decode_tps": 30.52,
      "e2e_s": 33.729,
      "finish_reason": "length",
      "server_decode_tps": 30.52
    },
    {
      "phase": "http-greedy",
      "prompt": 6,
      "prompt_tokens": 50,
      "completion_tokens": 1024,
      "ttft_ms": 199.66,
      "tpot_ms": 32.731,
      "decode_tps": 30.55,
      "e2e_s": 33.684,
      "finish_reason": "length",
      "server_decode_tps": 30.55
    },
    {
      "phase": "http-greedy",
      "prompt": 7,
      "prompt_tokens": 240,
      "completion_tokens": 419,
      "ttft_ms": 306.78,
      "tpot_ms": 32.686,
      "decode_tps": 30.59,
      "e2e_s": 14.003,
      "finish_reason": "stop",
      "server_decode_tps": 30.52
    }
  ],
  "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/64k-q8kv",
      "--fit",
      "off",
      "-c",
      "65536",
      "-np",
      "1",
      "--cache-ram",
      "0",
      "-ngl",
      "999",
      "-fa",
      "on",
      "-ctk",
      "q8_0",
      "-ctv",
      "q8_0",
      "-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": {}
  }
}