4 × NVIDIA RTX PRO 6000 Blackwell Server Edition

Tested system

Processor: AMD EPYC 9124
Graphics: NVIDIA RTX PRO 6000 Blackwell Server Edition
Accelerator count: 4
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 / llama3.1-8bOffline · v6.1 · Tokens/s

4 × NVIDIA RTX PRO 6000 Blackwell Server Edition

24,511.80^* Tokens/s

Accuracy target / result: ROUGE1: 38.5238 ROUGE2: 15.9869 ROUGEL: 24.3988 ROUGELSUM: 35.6579 GEN_LEN: 8178576
Weight precision: fp4
Division: closed
System type: datacenter
Availability: available
Cores per processor: 16
Processors per node: 1
Nodes: 1
Total accelerators: 4
Software stack: TensorRT 10.14.1.48, CUDA 13.1
Operating system: Rocky Linux 9.7

MLPerf Inference / llama3.1-8bServer · v6.1 · Tokens/s

4 × NVIDIA RTX PRO 6000 Blackwell Server Edition

24,510.50^* Tokens/s

Accuracy target / result: ROUGE1: 38.5242 ROUGE2: 15.9887 ROUGEL: 24.3995 ROUGELSUM: 35.6582 GEN_LEN: 8178682
Weight precision: fp4
Division: closed
System type: datacenter
Availability: available
Cores per processor: 16
Processors per node: 1
Nodes: 1
Total accelerators: 4
Software stack: TensorRT 10.14.1.48, CUDA 13.1
Operating system: Rocky Linux 9.7

Processing & platform

Processor: AMD EPYC 9124
Graphics: NVIDIA RTX PRO 6000 Blackwell Server Edition

Connectivity & software

Operating system: Rocky Linux 9.7

System configuration

Reported host core count: 16
Processors per node: 1
Accelerator count: 4
Nodes: 1
Software stack: TensorRT 10.14.1.48, CUDA 13.1
Submission division: closed

* Provisional data