LATEST INTELLIGENCE
AI REFERENCE DESIGN
1 SU NVIDIA DGX H100 SUPERPOD
PRESENTED BY
Legrand, a global specialist in electrical and digital building infrastructures, has been a leading innovator in the data center industry for over three decades. With the rise of artificial intelligence( AI), Legrand created a comprehensive AI Reference Design aimed at supporting the generative AI systems going into modern data centers. This design utilizes technologies that improve the functionality, efficiency, and user experience of today’ s data center infrastructure by leveraging our experience and expertise in the areas of power, thermal management, and connectivity for the data center. It enables sophisticated machine learning algorithms and data analytics to take place on data gathered from our sensors and power products. By continuously monitoring and analyzing temperature, humidity, and power data, the data center operator can make real-time adjustments to cooling and server workloads that optimize performance and energy consumption. This leads to significant cost savings opportunities and supports sustainability efforts.
The Legrand AI Reference Design supports the implementation of NVIDIA DGX H100 SuperPODs.
About the Nvidia DGX superpod
• A DGX SuperPOD is built up from smaller modules referred to as Scalable Units( SU).
• A full SuperPOD is comprised of four Scalable Units( 4 SU).
• Each SU contains 31 or 32 DGX nodes, with 4 DGX H100 systems per rack across
20 INTELLIGENTCIO NORTH AMERICA www. intelligentcio. com