Integrating Composable Memory Solutions into Storage-Intensive Workloads
Open Compute Project via YouTube
Overview
Watch a technical presentation from the Open Compute Project exploring solutions for memory-intensive applications in high-performance computing and cloud environments. Learn how data centers can address challenges of achieving superior performance and lower TCO through CXL memory expansion and pooling approaches. Examine specific analysis of memory pooling implementation for SPARK workloads, with experts from Ampere Computing, Samsung, and Micron discussing techniques to reduce node count and memory footprint while avoiding issues like increased TCO, signal integrity problems, and reduced bandwidth that come with traditional solutions like 2DPC and 3D stacking.
Syllabus
Integrating Composable Memory Solutions into Storage intensive workloads
Taught by
Open Compute Project