{
  "backend": "vllm-omni",
  "task": "ref2va",
  "gpus_per_instance": 4,
  "dit_tensor_parallel_size": 2,
  "ulysses_degree": 2,
  "replicas": 2,
  "expected_requests": 32,
  "requests_recorded": 32,
  "completed": 32,
  "failed": 0,
  "machine_wall_s": 4499.805894613266,
  "machine_qps": 0.0071114178587808234,
  "latency_mean_s": 281.0816363893391,
  "latency_p50_s": 222.61875976994634,
  "latency_p95_s": 573.1763278442552,
  "by_short_edge": {
    "480": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 106.20279711237526,
      "latency_p95_s": 106.431321673817
    },
    "720": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 212.077966982848,
      "latency_p95_s": 212.1960737178917
    },
    "768": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 233.23018924598,
      "latency_p95_s": 233.52808195799588
    },
    "1080": {
      "requests": 8,
      "completed": 8,
      "failed": 0,
      "latency_mean_s": 572.8155922161532,
      "latency_p95_s": 574.0361641392926
    }
  }
}