Learn about evolving memory requirements for AI systems in hyperscale data centers in this 33-minute technical presentation from Meta's AI Systems Technologist Manoj Wadekar and Microsoft's Principal Architect Samir Rajadnya. Explore how data center infrastructure is shifting from traditional CPU-centric platforms optimized for scale-out stateless applications to GPU/accelerator-focused systems that support next-generation AI applications. Discover the innovations needed to address growing memory demands as hyperscale facilities adapt to support increasingly complex artificial intelligence workloads and zettabyte-scale storage requirements.
Memory Expansion Requirements for AI Systems in Hyperscale Data Centers
Open Compute Project via YouTube
Overview
Syllabus
General compute and AI Needs for Memory Expansion for Hyperscalers
Taught by
Open Compute Project
Related Courses
-
Pushing Infrastructure Boundaries: Memory Requirements in AI Data Centers
-
Infrastructure Software for Hyperscale AI and GPU Clusters
-
Data Center Infrastructure Design - An Introduction
-
Chiplet-Based Compressed LLC Cache and Memory Expansion for Data Centers
-
Optical Interconnects for Large-Scale AI Clusters - A Meta Perspective
-
Unleashing the Full Potential of CXL Memory Technology in Data Centers
