Strengthened AI Infrastructure for the New Era: From Systems to Liquid-Cooled Clusters (Presented by Supermicro)
The rise of AI has led to businesses expanding the data center infrastructure that supports it. In this new frontier, traditional IT hardware principles must be re-evaluated as AI computing presents unique challenges that lead to increasing complexity and power consumption while scaling deployment size. On the system-level, Supermicro reveals insight on the design processes behind new design approaches to maximize thermals, power delivery, and performance of modern AI CPUs and GPUs. Building up to the rack level, Supermicro dives in to best-practices for AI cluster design and network topology, as well as making the case for direct-to-chip liquid cooling as the logical choice for cooling modern scale-out AI datacenters.