The Case for Network Coding for Collective Communication on HPC Interconnection Networks

Abstract

SUMMARYRecently network bandwidth becomes a performance concern particularly for collective communication since bisection bandwidths of supercomputers become far less than their full bisection bandwidths. In this context we propose the use of a network coding technique to reduce the number of unicasts and the size of data transferred in latency-sensitive collective communications in supercomputers. Our proposed network coding scheme has a hierarchical multicasting structure with intra-group and inter-group unicasts. Quantitative analysis show that the aggregate path hop counts by our hierarchical network coding decrease as much as 94% when compared to conventional unicast-based multicasts. We validate these results by cycle-accurate network simulations. In 1,024-switch networks, the network reduces the execution time of collective communications as much as 70%. We also show that our hierarchical network coding is beneficial for any packet size.

NEST: Towards Extreme Scale Computing Systems

“…Demand for computing power in these fields impose great pressure on the network size of high-performance computing (HPC) systems. With the extraordinary growth in the number of computing nodes, performance is increasingly determined by how data is transmitted among the numerous compute resources [1], [2]. The efficient interconnection of hundreds of thousands of computing elements is proved to be one of the most critical barriers toward realizing exascale systems [3].…”