{
  "285": {
    "Organization": "NVIDIA",
    "Availability": "available",
    "Division": "closed",
    "SystemType": "datacenter",
    "SystemName": "NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT)",
    "Platform": "B300-SXM-270GBx8_TRT",
    "Model": "deepseek-r1",
    "MlperfModel": "deepseek-r1",
    "Scenario": "Offline",
    "Result": 70132.3,
    "Accuracy": "exact_match: 80.78851412944394  TOKENS_PER_SAMPLE: 3793.9359617137648",
    "number_of_nodes": 1,
    "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8570",
    "host_processors_per_node": 2,
    "host_processor_core_count": "56",
    "accelerator_model_name": "NVIDIA B300-SXM-270GB",
    "accelerators_per_node": 8,
    "total_accelerators": 8,
    "Location": "./closed/NVIDIA/results/B300-SXM-270GBx8_TRT/deepseek-r1/Offline/performance/run_1",
    "framework": "TensorRT 10.14, CUDA 13.1, cuDNN 9.17, TensorRT-LLM feat/1.2-mlpinf, NVIDIA Dynamo mlperf-v6.0-dynamo-v0.8.0, vLLM CentML:mlperf-inf-mm-q3vl-v6.1",
    "operating_system": "Ubuntu 24.04",
    "notes": "B300 TGP 1100W",
    "compliance": "closed",
    "errors": 0,
    "version": "v6.1",
    "inferred": 0,
    "has_power": false,
    "Units": "Tokens/s",
    "weight_data_types": "fp4",
    "design_power_watts": null,
    "PrivateID": null
  },
  "286": {
    "Organization": "NVIDIA",
    "Availability": "available",
    "Division": "closed",
    "SystemType": "datacenter",
    "SystemName": "NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT)",
    "Platform": "B300-SXM-270GBx8_TRT",
    "Model": "deepseek-r1",
    "MlperfModel": "deepseek-r1",
    "Scenario": "Server",
    "Result": 59867.8,
    "Accuracy": "exact_match: 81.17593436645396  TOKENS_PER_SAMPLE: 3818.7203737465816",
    "number_of_nodes": 1,
    "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8570",
    "host_processors_per_node": 2,
    "host_processor_core_count": "56",
    "accelerator_model_name": "NVIDIA B300-SXM-270GB",
    "accelerators_per_node": 8,
    "total_accelerators": 8,
    "Location": "./closed/NVIDIA/results/B300-SXM-270GBx8_TRT/deepseek-r1/Server/performance/run_1",
    "framework": "TensorRT 10.14, CUDA 13.1, cuDNN 9.17, TensorRT-LLM feat/1.2-mlpinf, NVIDIA Dynamo mlperf-v6.0-dynamo-v0.8.0, vLLM CentML:mlperf-inf-mm-q3vl-v6.1",
    "operating_system": "Ubuntu 24.04",
    "notes": "B300 TGP 1100W",
    "compliance": "closed",
    "errors": 0,
    "version": "v6.1",
    "inferred": 0,
    "has_power": false,
    "Units": "Tokens/s",
    "weight_data_types": "fp4",
    "design_power_watts": null,
    "PrivateID": null
  },
  "287": {
    "Organization": "NVIDIA",
    "Availability": "available",
    "Division": "closed",
    "SystemType": "datacenter",
    "SystemName": "NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT)",
    "Platform": "B300-SXM-270GBx8_TRT",
    "Model": "gpt-oss-120b",
    "MlperfModel": "gpt-oss-120b",
    "Scenario": "Offline",
    "Result": 106637.0,
    "Accuracy": "exact_match: 83.430",
    "number_of_nodes": 1,
    "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8570",
    "host_processors_per_node": 2,
    "host_processor_core_count": "56",
    "accelerator_model_name": "NVIDIA B300-SXM-270GB",
    "accelerators_per_node": 8,
    "total_accelerators": 8,
    "Location": "./closed/NVIDIA/results/B300-SXM-270GBx8_TRT/gpt-oss-120b/Offline/performance/run_1",
    "framework": "TensorRT 10.14, CUDA 13.1, cuDNN 9.17, TensorRT-LLM feat/1.2-mlpinf, NVIDIA Dynamo mlperf-v6.0-dynamo-v0.8.0, vLLM CentML:mlperf-inf-mm-q3vl-v6.1",
    "operating_system": "Ubuntu 24.04",
    "notes": "B300 TGP 1100W",
    "compliance": "closed",
    "errors": 0,
    "version": "v6.1",
    "inferred": 0,
    "has_power": false,
    "Units": "Tokens/s",
    "weight_data_types": "fp4",
    "design_power_watts": null,
    "PrivateID": null
  },
  "288": {
    "Organization": "NVIDIA",
    "Availability": "available",
    "Division": "closed",
    "SystemType": "datacenter",
    "SystemName": "NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT)",
    "Platform": "B300-SXM-270GBx8_TRT",
    "Model": "gpt-oss-120b",
    "MlperfModel": "gpt-oss-120b",
    "Scenario": "Server",
    "Result": 100630.0,
    "Accuracy": "exact_match: 83.702",
    "number_of_nodes": 1,
    "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8570",
    "host_processors_per_node": 2,
    "host_processor_core_count": "56",
    "accelerator_model_name": "NVIDIA B300-SXM-270GB",
    "accelerators_per_node": 8,
    "total_accelerators": 8,
    "Location": "./closed/NVIDIA/results/B300-SXM-270GBx8_TRT/gpt-oss-120b/Server/performance/run_1",
    "framework": "TensorRT 10.14, CUDA 13.1, cuDNN 9.17, TensorRT-LLM feat/1.2-mlpinf, NVIDIA Dynamo mlperf-v6.0-dynamo-v0.8.0, vLLM CentML:mlperf-inf-mm-q3vl-v6.1",
    "operating_system": "Ubuntu 24.04",
    "notes": "B300 TGP 1100W",
    "compliance": "closed",
    "errors": 0,
    "version": "v6.1",
    "inferred": 0,
    "has_power": false,
    "Units": "Tokens/s",
    "weight_data_types": "fp4",
    "design_power_watts": null,
    "PrivateID": null
  },
  "289": {
    "Organization": "NVIDIA",
    "Availability": "available",
    "Division": "closed",
    "SystemType": "datacenter",
    "SystemName": "NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT)",
    "Platform": "B300-SXM-270GBx8_TRT",
    "Model": "llama2-70b-99.9",
    "MlperfModel": "llama2-70b-99.9",
    "Scenario": "Offline",
    "Result": 109277.0,
    "Accuracy": "ROUGE1: 44.7951  ROUGE2: 22.3759  ROUGEL: 29.1534  TOKENS_PER_SAMPLE: 272.8",
    "number_of_nodes": 1,
    "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8570",
    "host_processors_per_node": 2,
    "host_processor_core_count": "56",
    "accelerator_model_name": "NVIDIA B300-SXM-270GB",
    "accelerators_per_node": 8,
    "total_accelerators": 8,
    "Location": "./closed/NVIDIA/results/B300-SXM-270GBx8_TRT/llama2-70b-99.9/Offline/performance/run_1",
    "framework": "TensorRT 10.14, CUDA 13.1, cuDNN 9.17, TensorRT-LLM feat/1.2-mlpinf, NVIDIA Dynamo mlperf-v6.0-dynamo-v0.8.0, vLLM CentML:mlperf-inf-mm-q3vl-v6.1",
    "operating_system": "Ubuntu 24.04",
    "notes": "B300 TGP 1100W",
    "compliance": "closed",
    "errors": 0,
    "version": "v6.1",
    "inferred": 0,
    "has_power": false,
    "Units": "Tokens/s",
    "weight_data_types": "fp4",
    "design_power_watts": null,
    "PrivateID": null
  },
  "290": {
    "Organization": "NVIDIA",
    "Availability": "available",
    "Division": "closed",
    "SystemType": "datacenter",
    "SystemName": "NVIDIA DGX B300 (8x B300-SXM-270GB, TensorRT)",
    "Platform": "B300-SXM-270GBx8_TRT",
    "Model": "llama2-70b-99.9",
    "MlperfModel": "llama2-70b-99.9",
    "Scenario": "Server",
    "Result": 106949.0,
    "Accuracy": "ROUGE1: 44.8452  ROUGE2: 22.4177  ROUGEL: 29.2149  TOKENS_PER_SAMPLE: 272.7",
    "number_of_nodes": 1,
    "host_processor_model_name": "Intel(R) Xeon(R) Platinum 8570",
    "host_processors_per_node": 2,
    "host_processor_core_count": "56",
    "accelerator_model_name": "NVIDIA B300-SXM-270GB",
    "accelerators_per_node": 8,
    "total_accelerators": 8,
    "Location": "./closed/NVIDIA/results/B300-SXM-270GBx8_TRT/llama2-70b-99.9/Server/performance/run_1",
    "framework": "TensorRT 10.14, CUDA 13.1, cuDNN 9.17, TensorRT-LLM feat/1.2-mlpinf, NVIDIA Dynamo mlperf-v6.0-dynamo-v0.8.0, vLLM CentML:mlperf-inf-mm-q3vl-v6.1",
    "operating_system": "Ubuntu 24.04",
    "notes": "B300 TGP 1100W",
    "compliance": "closed",
    "errors": 0,
    "version": "v6.1",
    "inferred": 0,
    "has_power": false,
    "Units": "Tokens/s",
    "weight_data_types": "fp4",
    "design_power_watts": null,
    "PrivateID": null
  },
  "300": {
    "Organization": "NVIDIA",
    "Availability": "available",
    "Division": "closed",
    "SystemType": "datacenter",
    "SystemName": "NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT)",
    "Platform": "GB200-NVL72_GB200-186GB_aarch64x72_TRT",
    "Model": "gpt-oss-120b",
    "MlperfModel": "gpt-oss-120b",
    "Scenario": "Offline",
    "Result": 883761.0,
    "Accuracy": "exact_match: 84.038",
    "number_of_nodes": 18,
    "host_processor_model_name": "NVIDIA Grace CPU",
    "host_processors_per_node": 2,
    "host_processor_core_count": "144",
    "accelerator_model_name": "NVIDIA GB200",
    "accelerators_per_node": 4,
    "total_accelerators": 72,
    "Location": "./closed/NVIDIA/results/GB200-NVL72_GB200-186GB_aarch64x72_TRT/gpt-oss-120b/Offline/performance/run_1",
    "framework": "TensorRT 10.14, CUDA 13.1, cuDNN 9.17, TensorRT-LLM feat/1.2-mlpinf, NVIDIA Dynamo mlperf-v6.0-dynamo-v0.8.0, vLLM CentML:mlperf-inf-mm-q3vl-v6.1",
    "operating_system": "Ubuntu 24.04",
    "notes": "NVIDIA GB200 NVL72, 1200W TGP Per GPU",
    "compliance": "closed",
    "errors": 0,
    "version": "v6.1",
    "inferred": 0,
    "has_power": false,
    "Units": "Tokens/s",
    "weight_data_types": "fp4",
    "design_power_watts": null,
    "PrivateID": null
  },
  "301": {
    "Organization": "NVIDIA",
    "Availability": "available",
    "Division": "closed",
    "SystemType": "datacenter",
    "SystemName": "NVIDIA GB200 NVL72 (72x GB200-186GB_aarch64, TensorRT)",
    "Platform": "GB200-NVL72_GB200-186GB_aarch64x72_TRT",
    "Model": "gpt-oss-120b",
    "MlperfModel": "gpt-oss-120b",
    "Scenario": "Server",
    "Result": 899005.0,
    "Accuracy": "exact_match: 83.914",
    "number_of_nodes": 18,
    "host_processor_model_name": "NVIDIA Grace CPU",
    "host_processors_per_node": 2,
    "host_processor_core_count": "144",
    "accelerator_model_name": "NVIDIA GB200",
    "accelerators_per_node": 4,
    "total_accelerators": 72,
    "Location": "./closed/NVIDIA/results/GB200-NVL72_GB200-186GB_aarch64x72_TRT/gpt-oss-120b/Server/performance/run_1",
    "framework": "TensorRT 10.14, CUDA 13.1, cuDNN 9.17, TensorRT-LLM feat/1.2-mlpinf, NVIDIA Dynamo mlperf-v6.0-dynamo-v0.8.0, vLLM CentML:mlperf-inf-mm-q3vl-v6.1",
    "operating_system": "Ubuntu 24.04",
    "notes": "NVIDIA GB200 NVL72, 1200W TGP Per GPU",
    "compliance": "closed",
    "errors": 0,
    "version": "v6.1",
    "inferred": 0,
    "has_power": false,
    "Units": "Tokens/s",
    "weight_data_types": "fp4",
    "design_power_watts": null,
    "PrivateID": null
  }
}