{
  "backend": "vllm-omni",
  "task": "ref2va",
  "gpus_per_instance": 2,
  "dit_tensor_parallel_size": 2,
  "ulysses_degree": 1,
  "replicas": 4,
  "expected_requests": 32,
  "requests_recorded": 32,
  "completed": 32,
  "failed": 0,
  "machine_wall_s": 4204.426504850388,
  "machine_qps": 0.007611026132359211,
  "latency_mean_s": 525.1911418869277,
  "latency_p50_s": 413.46523636602797,
  "latency_p95_s": 1084.747085970419,
  "by_short_edge": {
    "480": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 190.50694103751448,
      "latency_p95_s": 190.64961519601056
    },
    "720": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 391.9126197292353,
      "latency_p95_s": 392.6838105404866
    },
    "768": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 434.50199320795946,
      "latency_p95_s": 434.6102629141999
    },
    "1080": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 1083.8430135730014,
      "latency_p95_s": 1084.9118782492121
    }
  }
}