8 × NVIDIA RTX PRO 6000 Blackwell Server Edition

Tested system

Processor: Intel(R) Xeon(R) 6787P
Graphics: NVIDIA RTX PRO 6000 Blackwell Server Edition
Accelerator count: 8
Nodes: 1

Recorded results

Only the same benchmark, version, workload, scenario and unit share a row. Missing results are not zero. System results include the full submitted hardware and software configuration.

MLPerf Inference / llama2-70b-99.9Offline · v6.1 · Tokens/s

8 × NVIDIA RTX PRO 6000 Blackwell Server Edition

30,034.20^* Tokens/s

Accuracy target / result: ROUGE1: 44.7186 ROUGE2: 22.3366 ROUGEL: 29.1057 TOKENS_PER_SAMPLE: 275.5
Weight precision: fp4
Division: closed
System type: datacenter
Availability: available
Cores per processor: 172
Processors per node: 2
Nodes: 1
Total accelerators: 8
Software stack: TensorRT 10.14, CUDA 13.2
Operating system: Ubuntu 24.04.4

MLPerf Inference / llama2-70b-99.9Server · v6.1 · Tokens/s

8 × NVIDIA RTX PRO 6000 Blackwell Server Edition

29,203.30^* Tokens/s

Accuracy target / result: ROUGE1: 44.8047 ROUGE2: 22.3347 ROUGEL: 29.1315 TOKENS_PER_SAMPLE: 267.8
Weight precision: fp4
Division: closed
System type: datacenter
Availability: available
Cores per processor: 172
Processors per node: 2
Nodes: 1
Total accelerators: 8
Software stack: TensorRT 10.14, CUDA 13.2
Operating system: Ubuntu 24.04.4

Processing & platform

Processor: Intel(R) Xeon(R) 6787P
Graphics: NVIDIA RTX PRO 6000 Blackwell Server Edition

Connectivity & software

Operating system: Ubuntu 24.04.4

System configuration

Reported host core count: 172
Processors per node: 2
Accelerator count: 8
Nodes: 1
Software stack: TensorRT 10.14, CUDA 13.2
Submission division: closed

* Provisional data