HammingMesh: A Network Topology for Large-Scale Deep Learning.

This article presents HammingMesh, a flexible topology that overcomes current high-performance computing’s inability to support deep-learning workloads by allowing for the adjustment of the ratio of local and global bandwidth. The article discusses communication in distributed deep learning with a l...

Full description

Bibliographic Details

Published in:	Communications of the ACM Vol. 67; no. 12; pp. 97 - 106
Main Authors:	Hoefler, Torsten, Bonoto, Tommaso, De Sensi, Daniele, Di Girolamo, Salvatore, Li, Shigang, Heddes, Marco, Goel, Deepak, Castro, Miguel, Scott, Steve
Format:	Article
Published:	Association for Computing Machinery Dec2024
Subjects:	Deep learning High performance computing Computer networks Bandwidth allocation Artificial neural networks
Online Access:	View this record in EBSCOhost