Reports
Detailed configuration information are available to eligible subscribers at the links above. To learn more about subscription options, please contact us.
STAC recently audited a STAC-ML™ Markets (Inference) benchmark on a stack featuring an NVIDIA GH200 Grace Hopper Superchip in a Supermicro ARS-111GL-NHR server.
In this audit, the GH200 system was compared to a previous submission using FPGAs. The results show:
- Smallest model: Up to 20% lower latency
- Medium model: Up to 8% lower latency
- Largest model: 49% lower latency
Full benchmark reports are available at the link opposite to all members.
With additional performance, efficiency and quality results, detailed configurations, and code access available to premium subscribers.