Accelerating Real World CXL Deployments - From Hardware Assisted Intelligent Tiering to Compression
Open Compute Project via YouTube
Earn a Michigan Engineering AI Certificate — Stay Ahead of the AI Revolution
Earn Your Business Degree, Tuition-Free, 100% Online!
Overview
Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Explore two groundbreaking innovations that advance real-world CXL memory adoption through this technical conference presentation. Learn about the CXL Hot-range Monitoring Unit (CHMU), an in-device hardware engine that detects hot memory regions in real-time and enables intelligent tiering decisions like page migration, achieving up to 40% performance gains on AI workloads such as DLRM compared to CXL-only configurations. Discover a full-stack compressed memory system aligned with the OCP Hyperscale Tiered Memory Expander Spec that supports cacheline-level compression and runs real CMS workloads including Hyrise and HPCG under MGLRU-based tiering. Understand how these practical solutions translate CXL specifications into deployable systems, providing actionable insights for building dynamic, cost-effective tiered memory solutions at scale while demonstrating tangible improvements in dollar-per-GB efficiency and overall server performance.
Syllabus
Accelerating Real World CXL Deployments From Hardware Assisted Intelligent Tiering to Compres
Taught by
Open Compute Project