A Closer Look: MLPerf Inferencing Benchmarks: Showcasing Supermicro NVIDIA HGX™ B200 Systems

We recently had a webinar discussing the recent MLPerf inference testing results for Supermicro systems with the NVIDIA HGX B200 8-GPU. With leading performance in many categories we covered how you can use these results to help you make informed decisions on your AI infrastructure.

Webinar: MLPerf Inferencing Benchmarks: Showcasing Supermicro NVIDIA HGX™ B200 Systems

We featured Supermicro experts Alok Srivastav, Director, Solutions Management AI, Advait Italia, Systems Engineer, and Michael Schulman, Sr. Corporate Communications Manager.

Let’s look at some highlights and key takeaways from the webinar.

Key Takeaways:

• MLPerf benchmarks are crucial for understanding AI workloads.
• ML Commons provides a transparent testing environment for performance metrics.
• Supermicro offers a wide range of hardware options for AI applications.
• The B200 system shows significant performance improvements over previous generations.
• MLPerf results help customers make informed decisions about their infrastructure.
• Liquid cooling is becoming essential due to increasing power demands.
• The performance metrics include throughput, latency, and accuracy.
• The industry is moving towards more complex and powerful AI systems.
• Regular updates and submissions to MLPerf ensure ongoing relevance and accuracy.

Highlights:

MLPerf: Future-Proofing AI Infrastructure

Advait: “When you use these benchmarks and when you have these certifications from ML Commons, which is industry standard, you future-proof your infrastructure as well.”

Liquid Cooling: The Future of Data Centers

Alok: “Data centers are power limited. To get the best use of their power, the best option is the liquid cooling route.”

ML Commons: Ensuring Transparency in AI

Advait: “There has to be a level of transparency and a common ground where we all can agree and we all can conclude that this is a common level of benchmark”

These are just a few highlights. Watch the whole webinar to get more of an in-depth look at the benchmarks and take a deeper dive into the systems.

Additional Resources:

• GPU Servers
• Press release: Industry’s First-to-Market Supermicro NVIDIA HGX™ B200 Systems Demonstrate AI Performance Leadership on MLPerf® Inference v5.0 Results
• Supermicro AI Infrastructure Server Solutions
• Supermicro Webinar Channel