results: A100-SXM4-80GB Suite F with SGLang v6 runner

results/community/nvidia_a100_sxm4_80gbx1_suite_F_nvidia_sglang_inference_opt_4fd1e59e_1189951b/accuracy/accuracy.json

@@ -0,0 +1,8 @@
+{
+  "subset_score": 0.41,
+  "baseline_delta": 0.03,
+  "valid": true,
+  "framework": "SGLang",
+  "precision": "BF16",
+  "notes": "Integrated accuracy check \u2014 used same SGLang instance as benchmark."
+}

results/community/nvidia_a100_sxm4_80gbx1_suite_F_nvidia_sglang_inference_opt_4fd1e59e_1189951b/env_info.json

@@ -0,0 +1,64 @@
+{
+  "collected_at": "2026-07-31T02:37:50.313608+00:00",
+  "accelerators": [
+    {
+      "index": 5,
+      "name": "NVIDIA A100-SXM4-80GB",
+      "vendor": "NVIDIA",
+      "memory_gb": 80.0,
+      "driver_version": "570.133.20",
+      "firmware_version": null,
+      "compute_capability": "8.0",
+      "supports_bf16": true
+    },
+    {
+      "index": 7,
+      "name": "NVIDIA A100-SXM4-80GB",
+      "vendor": "NVIDIA",
+      "memory_gb": 80.0,
+      "driver_version": "570.133.20",
+      "firmware_version": null,
+      "compute_capability": "8.0",
+      "supports_bf16": true
+    }
+  ],
+  "accelerator_platform": "nvidia",
+  "accelerator_topology": null,
+  "intra_node_interconnect": null,
+  "cpu": {
+    "model": "Intel(R) Xeon(R) Processor @ 2.90GHz",
+    "physical_cores": 96,
+    "logical_cores": 96,
+    "numa_nodes": 2
+  },
+  "system_memory_gb": 1000.0,
+  "pcie_generation": "PCIe Gen 4",
+  "cpu_accelerator_bandwidth_gbs": null,
+  "network_interfaces": [
+    {
+      "name": "mlx5_0",
+      "type": "InfiniBand/RoCE",
+      "bandwidth_gbps": null
+    },
+    {
+      "name": "mlx5_1",
+      "type": "InfiniBand/RoCE",
+      "bandwidth_gbps": null
+    },
+    {
+      "name": "mlx5_2",
+      "type": "InfiniBand/RoCE",
+      "bandwidth_gbps": null
+    },
+    {
+      "name": "mlx5_3",
+      "type": "InfiniBand/RoCE",
+      "bandwidth_gbps": null
+    }
+  ],
+  "os": "Ubuntu 22.04.5 LTS",
+  "python_version": "3.11.13",
+  "kernel_version": "5.10.134-013.9.kangaroo.al8.x86_64",
+  "runtime_version": "CUDA 12.8",
+  "pytorch_version": "2.9.1+cu128"
+}

results/community/nvidia_a100_sxm4_80gbx1_suite_F_nvidia_sglang_inference_opt_4fd1e59e_1189951b/interactive/result.json

@@ -0,0 +1,169 @@
+{
+  "schema_version": "1.0",
+  "suite_id": "suite_F",
+  "implementation_id": "nvidia_sglang_inference_opt_4fd1e59e",
+  "chip": {
+    "name": "NVIDIA A100-SXM4-80GB",
+    "vendor": "NVIDIA",
+    "count": 1,
+    "memory_gb": 80.0,
+    "interconnect_intra_node": null,
+    "interconnect_inter_node": null
+  },
+  "environment": {
+    "collected_at": "2026-07-31T02:37:50.313608+00:00",
+    "accelerators": [
+      {
+        "index": 5,
+        "name": "NVIDIA A100-SXM4-80GB",
+        "vendor": "NVIDIA",
+        "memory_gb": 80.0,
+        "driver_version": "570.133.20",
+        "firmware_version": null,
+        "compute_capability": "8.0",
+        "supports_bf16": true
+      },
+      {
+        "index": 7,
+        "name": "NVIDIA A100-SXM4-80GB",
+        "vendor": "NVIDIA",
+        "memory_gb": 80.0,
+        "driver_version": "570.133.20",
+        "firmware_version": null,
+        "compute_capability": "8.0",
+        "supports_bf16": true
+      }
+    ],
+    "accelerator_platform": "nvidia",
+    "accelerator_topology": null,
+    "intra_node_interconnect": null,
+    "cpu": {
+      "model": "Intel(R) Xeon(R) Processor @ 2.90GHz",
+      "physical_cores": 96,
+      "logical_cores": 96,
+      "numa_nodes": 2
+    },
+    "system_memory_gb": 1000.0,
+    "pcie_generation": "PCIe Gen 4",
+    "cpu_accelerator_bandwidth_gbs": null,
+    "network_interfaces": [
+      {
+        "name": "mlx5_0",
+        "type": "InfiniBand/RoCE",
+        "bandwidth_gbps": null
+      },
+      {
+        "name": "mlx5_1",
+        "type": "InfiniBand/RoCE",
+        "bandwidth_gbps": null
+      },
+      {
+        "name": "mlx5_2",
+        "type": "InfiniBand/RoCE",
+        "bandwidth_gbps": null
+      },
+      {
+        "name": "mlx5_3",
+        "type": "InfiniBand/RoCE",
+        "bandwidth_gbps": null
+      }
+    ],
+    "os": "Ubuntu 22.04.5 LTS",
+    "python_version": "3.11.13",
+    "kernel_version": "5.10.134-013.9.kangaroo.al8.x86_64",
+    "runtime_version": "CUDA 12.8",
+    "pytorch_version": "2.9.1+cu128"
+  },
+  "software": {
+    "framework": "SGLang",
+    "framework_version": "0.5.6",
+    "driver_version": "570.133.20",
+    "runtime_version": "CUDA 12.8",
+    "os": "Ubuntu 22.04.5 LTS",
+    "python_version": "3.11.13"
+  },
+  "model": {
+    "model_id": "Qwen/Qwen2.5-0.5B-Instruct",
+    "model_revision": "7ae557604adf67be50417f59c2c2f167def9a775",
+    "model_name": null,
+    "model_note": "Local ModelScope snapshot for Suite F",
+    "model_source": "local",
+    "architecture": "dense",
+    "parameter_count_b": 0.5,
+    "precision": "BF16",
+    "effective_dtype": "bfloat16",
+    "quantization_method": null,
+    "model_format": "HuggingFace original"
+  },
+  "task": {
+    "scenario": "interactive",
+    "num_runs": 3,
+    "warmup_runs": 1,
+    "parallelism": {
+      "tensor_parallel_size": 1,
+      "pipeline_parallel_size": 1,
+      "expert_parallel_size": 1,
+      "data_parallel_size": 1
+    },
+    "extra_config": {
+      "enable_tokenizer_batch_encode": true,
+      "continuous_decode_steps": 4,
+      "engine_kwargs": {
+        "attention_backend": "triton",
+        "disable_cuda_graph_padding": true,
+        "enable_piecewise_cuda_graph": true,
+        "piecewise_cuda_graph_max_tokens": 1024,
+        "prefill_attention_backend": "fa3"
+      }
+    },
+    "runtime_metrics": null
+  },
+  "metrics": {
+    "interactive": {
+      "ttft_ms_p50": 7.19,
+      "ttft_ms_p90": 7.54,
+      "ttft_ms_p99": 15.06,
+      "tpot_ms_p50": 1.62,
+      "tpot_ms_p90": 1.64,
+      "tpot_ms_p99": 1.71,
+      "peak_memory_gb": null,
+      "elapsed_seconds_median": 46.2,
+      "ttft_ms_p99_reliability": {
+        "n": 3,
+        "mean": 23.42,
+        "std": 22.39,
+        "cv_pct": 95.62,
+        "stability": "high-variance",
+        "runs": [
+          49.2,
+          12.23,
+          8.83
+        ]
+      }
+    }
+  },
+  "accuracy": {
+    "subset_score": null,
+    "baseline_delta": null,
+    "valid": false,
+    "notes": "Run --scenario accuracy to check model accuracy."
+  },
+  "meta": {
+    "submitted_by": "ChiangYuhsin",
+    "submission_type": "individual",
+    "date": "2026-07-31",
+    "time": "02:46:54",
+    "run_id": "1189951b",
+    "run_name": "nvidia_a100_sxm4_80gbx1_suite_F_nvidia_sglang_inference_opt_4fd1e59e_1189951b",
+    "flagged": null,
+    "reproduce_script": "runners/nvidia_sglang_inference_opt_4fd1e59e/runner.py",
+    "env_info_file": "../env_info.json",
+    "log_file": "run.log",
+    "samples_file": "samples.jsonl",
+    "notes": null,
+    "benchmark_start_time": "2026-07-31T02:44:33.515731+00:00",
+    "benchmark_end_time": "2026-07-31T02:46:54.781708+00:00",
+    "benchmark_elapsed_minutes": 2.4,
+    "model_load_seconds": 47.7
+  }
+}