AI, Data Science & Cloud Certificates from Google, IBM & Meta
Get 20% off all career paths from fullstack to AI
Overview
AI, Data Science & Cloud Certificates from Google, IBM & Meta — 40% Off
One plan covers every Professional Certificate on Coursera. 40% off Coursera Plus Annual.
Unlock All Certificates
Explore the architecture and implementation of large-scale CXL memory expansion systems designed for 100 TiB-class workloads in this 18-minute conference presentation. Learn how CXL Consortium members Liqid and Micron are deploying novel CXL-based system architectures to maximize performance and capacity for demanding applications. Discover how these systems dramatically increase size limits for what fits in memory while enabling composable provisioning of large amounts of memory for both memory-expansion pooling and shared access scenarios. Understand how existing applications and workflows can achieve memory de-duplication through shared memory using the open-source fabric-attached memory file system (famfs), and see how these solutions enable data scaling without sharding for workloads that cannot be effectively sharded or quantized.
Syllabus
Large CXL Memory Expansion Systems and Workloads
Taught by
Open Compute Project