In-Network Collective Operations: Game Changer or Challenge for AI Workloads?
Torsten Hoefler, Mikhail Khalilov, Josiah Clark, Surendra Anubolu, Mohan Kalkunte, Karen Schramm, Eric Spada, Duncan Roweth, Keith Underwood, Adrian Caulfield, Abdul Kabbani, Amirreza Rastegari
TL;DR
The opportunities of in-network collective operations for accelerated collective operations in artificial intelligence (AI) workloads are summarized to make this important field accessible to nonexperts in AI or networking, fostering a connection between these communities.
Abstract
This paper summarizes the opportunities of in-network collective operations (INC) for accelerated collective operations in AI workloads. We provide sufficient detail to make this important field accessible to non-experts in AI or networking, fostering a connection between these communities. Consider two types of INC: Edge-INC, where the system is implemented at the node level, and Core-INC, where the system is embedded within network switches. We outline the potential performance benefits as well as six key obstacles in the context of both Edge-INC and Core-INC that may hinder their adoption. Finally, we present a set of predictions for the future development and application of INC.