Overview
Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about packet handling optimization and flow engineering techniques for dense AI Ethernet clusters in this 19-minute conference talk from the Open Compute Project. Explore background concepts in packet management within AI networking environments and discover best practice implementations available in open SONiC releases for fabric optimization. Compare industry-standard terminology with open-source alternatives to achieve maximum cluster performance. Examine Dynamic Load Balancing as an innovative approach that surpasses traditional networking methods, with practical insights from networking experts at Supermicro and Broadcom on engineering high-performance AI fabrics.
Syllabus
Analysis of flow engineering and balancing options on AI fabrics
Taught by
Open Compute Project