Accelerating CPU-based Distributed DNN Training on Modern HPC Clusters using BlueField-2 DPUs

Arpan Jain, Nawras Alnaasan, Aamir Shafi, Hari Subramoni, Dhabaleswar K. Panda 0001. Accelerating CPU-based Distributed DNN Training on Modern HPC Clusters using BlueField-2 DPUs. In IEEE Symposium on High-Performance Interconnects, HOTI 2021, Santa Clara, CA, USA, August 18-20, 2021. pages 17-24, IEEE, 2021. [doi]

Abstract

Abstract is missing.