{
  "build_id": "google/gemma-4-26b-a4b-it@awq-w4a16",
  "campaign": "2026-10-03-vitesse-en",
  "duplicate_of": null,
  "ended_at": "2026-10-03T19:05:34Z",
  "energy": {
    "flags": [
      "fr:solo_stream:counter-readings-disagree"
    ],
    "gpus_counted": 2,
    "passes": [
      "fr",
      "en",
      "fr2"
    ],
    "period_s": 1.0,
    "protocol": "nvml-energy/v1",
    "scope": "gpu-board",
    "source": "counter"
  },
  "engine": {
    "image": "vllm/vllm-openai:v0.29.0",
    "name": "vllm",
    "pre_pin": false,
    "version": "0.29.0"
  },
  "engine_args": {
    "gpu_memory_utilization": 0.9,
    "kv_cache_dtype": "bfloat16",
    "max_model_len": 16384,
    "max_num_batched_tokens": 8192,
    "max_num_seqs": 8,
    "prefix_caching": false
  },
  "evidence": [
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/energy-en.json",
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/energy-fr2.json",
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/energy.json",
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/gpu-snapshots.json",
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/launch.json",
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/pret-en-secondes.txt",
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/vitesse-en.json",
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/vitesse-fr2.json",
    "evidence/2026-10-03-vitesse-en/gemma4-26b-a4b-paire/vitesse.json"
  ],
  "hardware": "2x3090fe-nvlink",
  "kind": "speed",
  "metrics": {
    "agg_completion_tokens": 4096,
    "agg_concurrency": 8,
    "agg_energy_j": 2437.165,
    "agg_ok": 8,
    "agg_power_mean_w": 664.8,
    "agg_power_peak_w": 680.7,
    "agg_tok_per_j": 1.6806,
    "agg_tok_s": 1118.4,
    "agg_wall_s": 3.66,
    "en_agg_completion_tokens": 3823,
    "en_agg_concurrency": 8,
    "en_agg_energy_j": 2403.831,
    "en_agg_ok": 8,
    "en_agg_power_mean_w": 670.5,
    "en_agg_power_peak_w": 689.8,
    "en_agg_tok_per_j": 1.5904,
    "en_agg_tok_s": 1067.5,
    "en_agg_wall_s": 3.58,
    "en_p50_latency_s": 3.48,
    "en_solo_completion_tokens": 495,
    "en_solo_energy_j": 1539.021,
    "en_solo_power_mean_w": 600.8,
    "en_solo_power_peak_w": 617.9,
    "en_solo_tok_per_j": 0.3216,
    "en_solo_tok_s": 193.6,
    "en_solo_ttft_ms": 71,
    "en_solo_wall_s": 2.557,
    "fr2_agg_completion_tokens": 4096,
    "fr2_agg_concurrency": 8,
    "fr2_agg_energy_j": 2421.492,
    "fr2_agg_ok": 8,
    "fr2_agg_power_mean_w": 665.3,
    "fr2_agg_power_peak_w": 689.0,
    "fr2_agg_tok_per_j": 1.6915,
    "fr2_agg_tok_s": 1127.9,
    "fr2_agg_wall_s": 3.63,
    "fr2_p50_latency_s": 3.63,
    "fr2_solo_completion_tokens": 512,
    "fr2_solo_energy_j": 1627.259,
    "fr2_solo_power_mean_w": 614.9,
    "fr2_solo_power_peak_w": 625.8,
    "fr2_solo_tok_per_j": 0.3146,
    "fr2_solo_tok_s": 193.8,
    "fr2_solo_ttft_ms": 73,
    "fr2_solo_wall_s": 2.642,
    "p50_latency_s": 3.66,
    "ready_s": 176,
    "solo_completion_tokens": 512,
    "solo_energy_j": 1592.899,
    "solo_power_mean_w": 600.4,
    "solo_power_peak_w": 613.2,
    "solo_tok_per_j": 0.3214,
    "solo_tok_s": 193.4,
    "solo_ttft_ms": 128,
    "solo_wall_s": 2.647
  },
  "model_id": "google/gemma-4-26b-a4b-it",
  "parse_notes": [
    "after-load snapshot taken 5 s after ready; no quality bench ran before the speed passes",
    "energy (nvml-energy/v1): NVIDIA's cumulative energy counter of the model's 2 cards, read at the start and end of each timed part, power read once a second; GPU boards only, summed, not the host CPU, RAM or power supplies; the streamed single request is in energy*.json only",
    "energy flags: fr:solo_stream:counter-readings-disagree; counter-readings-disagree means the counter and the integral of the 1-s power readings differ by more than 15 % on at least one card, and the counter's figure is kept",
    "energy metrics per pass (no prefix, en_, fr2_): solo_* and agg_* energy_j, power_mean_w, power_peak_w (the highest 1-s power reading, summed over the cards) and tok_per_j; agg_completion_tokens is the eight answers' tokens",
    "launch line of run 2026-09-25-spec-ab/g26-base-Aprime-tp2",
    "one discarded warm-up request (at most 8 tokens) was sent before the speed passes",
    "three speed passes in one session on the same server: French (headline keys, vitesse.json), the English twin (en_* keys, protocol speed-house/v1-en, vitesse-en.json), French again (fr2_* keys, vitesse-fr2.json: the session's drift, not a repetition)"
  ],
  "prompt_lang": "fr",
  "protocol": "speed-house/v1",
  "quality": [],
  "run_id": "2026-10-03-vitesse-en/gemma4-26b-a4b-paire",
  "scrubbed_fields": [
    "energy-en.json:candidate",
    "energy-en.json:driver_version",
    "energy-en.json:gpu_choice",
    "energy-en.json:gpus[].uuid",
    "energy-en.json:power_and_memory_of_every_card_seen",
    "energy-fr2.json:candidate",
    "energy-fr2.json:driver_version",
    "energy-fr2.json:gpu_choice",
    "energy-fr2.json:gpus[].uuid",
    "energy-fr2.json:power_and_memory_of_every_card_seen",
    "energy.json:candidate",
    "energy.json:driver_version",
    "energy.json:gpu_choice",
    "energy.json:gpus[].uuid",
    "energy.json:power_and_memory_of_every_card_seen",
    "launch.container_name",
    "launch.host",
    "launch.labels",
    "launch.model_path",
    "launch.port",
    "launch.ports",
    "launch.volumes",
    "vitesse-en.json:prompt_sha256",
    "vitesse-fr2.json:prompt_sha256"
  ],
  "source": {
    "source_id": "src-7328c18cc4b40546"
  },
  "started_at": "2026-10-03T19:05:07Z",
  "status": "ok",
  "topology": {
    "cpu_offload_gb": 0,
    "expert_parallel": false,
    "gpus": 2,
    "nvlink_used": true,
    "p2p": "nvlink",
    "pp": null,
    "ram_spill": false,
    "split": "tensor",
    "tp": 2
  },
  "vram": {
    "kind": "reserved",
    "mib_per_gpu": [
      21967,
      21967
    ],
    "snapshot": "load"
  }
}