18:48
Homa: The End of TCP for AI Clusters — John Ousterhout, Stanford
John Ousterhout
AI inference networking · GPU synchronization and tail latency · Incast
← All topics
AI Engineer topic
AI inference networking
1 AI Engineer conference talks about AI inference networking.