Files

115 lines
2.3 KiB
JSON

{
"case": {
"id": "08-latest-nvfp4-72k-embedded-mtp3-p010",
"model": "nvfp4-latest",
"context": 72000,
"split": [
72,
28
],
"projector": "off",
"mtp": true,
"mtp_max": 3,
"quality": false,
"long_fill": 0.0,
"vision": false
},
"model_path": "/models/qwen3.8-nvfp4-split/Qwen3.8-27B-NVFP4-Q4_K_M-mtp.gguf",
"started": "2026-08-22T21:21:49.422968+00:00",
"command": [
"docker",
"run",
"-d",
"--name",
"mike-ai-qwen38-dualgpu-bench",
"--gpus",
"all",
"--ipc",
"host",
"--read-only",
"--tmpfs",
"/tmp:size=1g,mode=1777",
"--security-opt",
"no-new-privileges:true",
"--cap-drop",
"ALL",
"-e",
"NVIDIA_VISIBLE_DEVICES=GPU-8ad38c6c-5a01-9d8e-1dfa-ed662ad78fbe,GPU-4834d9d7-5b61-3004-1fb3-4ae49d482d4b",
"-v",
"/data/models:/models:ro",
"-p",
"127.0.0.1:18099:8080",
"mike-ai/llama.cpp:3f545bec-test",
"--model",
"/models/qwen3.8-nvfp4-split/Qwen3.8-27B-NVFP4-Q4_K_M-mtp.gguf",
"--alias",
"08-latest-nvfp4-72k-embedded-mtp3-p010",
"--ctx-size",
"72000",
"--flash-attn",
"on",
"--cache-type-k",
"q4_0",
"--cache-type-v",
"q4_0",
"--threads",
"6",
"--threads-batch",
"6",
"--batch-size",
"64",
"--ubatch-size",
"32",
"--parallel",
"1",
"--jinja",
"--reasoning",
"auto",
"--reasoning-budget",
"768",
"--host",
"0.0.0.0",
"--port",
"8080",
"--metrics",
"--fit",
"off",
"--n-gpu-layers",
"all",
"--no-mmap",
"--no-ui",
"--temperature",
"0.2",
"--top-p",
"0.8",
"--top-k",
"20",
"--device",
"CUDA0,CUDA1",
"--main-gpu",
"0",
"--split-mode",
"layer",
"--tensor-split",
"72,28",
"--spec-type",
"draft-mtp",
"--spec-draft-n-max",
"3",
"--spec-draft-type-k",
"f16",
"--spec-draft-type-v",
"f16",
"--spec-draft-p-min",
"0.10"
],
"docker_launch_returncode": 0,
"docker_launch_stderr": "",
"loaded": false,
"load_seconds": 8.099,
"status": "load_failed",
"server_log": "/data/benchmarks/qwen38-final-acceptance-20260822/cases/08-latest-nvfp4-72k-embedded-mtp3-p010/server.log",
"oom": false,
"finished": "2026-08-22T21:21:57.880947+00:00"
}