{
  "backend": "vllm-omni",
  "task": "ref2va",
  "gpus_per_instance": 8,
  "dit_tensor_parallel_size": 2,
  "ulysses_degree": 4,
  "replicas": 1,
  "expected_requests": 32,
  "requests_recorded": 32,
  "completed": 32,
  "failed": 0,
  "machine_wall_s": 4734.41522359848,
  "machine_qps": 0.006759018482472225,
  "latency_mean_s": 147.9499400610075,
  "latency_p50_s": 119.12891493900679,
  "latency_p95_s": 295.50401763169793,
  "by_short_edge": {
    "480": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 58.23869467526674,
      "latency_p95_s": 58.36102275213925
    },
    "720": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 113.5257983354968,
      "latency_p95_s": 113.72436507834354
    },
    "768": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 124.60925887679332,
      "latency_p95_s": 124.76920832397882
    },
    "1080": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 295.4260083564732,
      "latency_p95_s": 296.4830407598987
    }
  }
}