SC20 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

Achieving Computation-Communication Overlap with Overdecomposition on GPU Systems


Workshop:ESPM2 2020: Fifth International Workshop on Extreme Scale Programming Models and Middleware

Authors: Jaemin Choi (University of Illinois), David F. Richards (Lawrence Livermore National Laboratory), and Laxmikant V. Kale (University of Illinois)


Abstract: The landscape of high performance computing is shifting towards a collection of multi-GPU nodes, widening the gap between on-node compute and off-node communication capabilities. Consequently, the ability to tolerate communication latencies and maximize utilization of the compute hardware are becoming increasingly important in achieving high performance. Overdecomposition has been successfully adopted on traditional CPU-based systems to achieve computation-communication overlap, significantly reducing the impact of communication on application performance. However, it has been unclear whether overdecomposition can provide the same benefits on modern GPU systems. In this work, we address the challenges in achieving computation-communication overlap with overdecomposition on GPU systems using the Charm++ parallel programming system. By prioritizing communication with CUDA streams in the application and supporting asynchronous progress of GPU operations in the Charm++ runtime system, we obtain improvements in overall performance of up to 50% and 47% with proxy applications Jacobi3D and MiniMD, respectively.





Back to ESPM2 2020: Fifth International Workshop on Extreme Scale Programming Models and Middleware Archive Listing



Back to Full Workshop Archive Listing