Watch a 14-minute conference talk exploring how Composable Disaggregated Infrastructure (CDI) introduces revolutionary distributed system architecture for dynamic compute systems. Learn how traditional architectures fall short for modern AI/ML tasks and discover how high-performance, multi-path network fabric design enables PCIe extension beyond compute nodes and racks. Presented by Matthew Williams from Rockport Networks at the Open Compute Project, explore the potential of configurable, efficient row-scale computing solutions that will power future AI/ML model training and large-scale inference operations in tomorrow's data centers.
Overview
Syllabus
AI/HPC: The Future of AI/ML Innovation is Row-Scale Disaggregation
Taught by
Open Compute Project
Related Courses
-
Monitoring GPUs at Scale for AI - ML and HPC Clusters
-
Accelerating AI and ML: Architectures and Systems for Efficient Computing
-
Scheduled Ethernet Fabric for Large-Scale AI Training Clusters
-
PCIe 6.0 Specification and Beyond - Enabling Storage for Machine Learning Applications
-
Future of AI Networks - Understanding UALink and Ultra Ethernet Consortium Solutions
-
Delivering Unprecedented Scale for AI Infrastructure and Cloud Connectivity
