Using DCT-based Approximate Communication to Improve MPI Performance in Parallel Clusters

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Communication overheads in distributed systems constitute a large fraction of the total execution time, and limit the scalability of applications running on these systems. We propose a DCT-based approximate communication scheme that takes advantage of the error resiliency of several widely-used applications, and improves communication efficiency by substantially reducing message lengths. Our scheme is implemented into the Message Passing Interface (MPI) library. When evaluated on several representative MPI applications on a real cluster system, it is seen that the fraction of total execution time devoted to communication reduces from 59% to 23%, even accounting for the computational overhead required for DCT encoding. For many communication-intensive applications, it is shown that our approximate communication scheme effectively speeds up the total execution time without much loss in quality of the result.

Original languageEnglish (US)
Title of host publication2019 IEEE 38th International Performance Computing and Communications Conference, IPCCC 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728110257
DOIs
StatePublished - Oct 2019
Event38th IEEE International Performance Computing and Communications Conference, IPCCC 2019 - London, United Kingdom
Duration: Oct 29 2019Oct 31 2019

Publication series

Name2019 IEEE 38th International Performance Computing and Communications Conference, IPCCC 2019

Conference

Conference38th IEEE International Performance Computing and Communications Conference, IPCCC 2019
CountryUnited Kingdom
CityLondon
Period10/29/1910/31/19

    Fingerprint

Cite this

Fan, Q., Lilja, D. J., & Sapatnekar, S. S. (2019). Using DCT-based Approximate Communication to Improve MPI Performance in Parallel Clusters. In 2019 IEEE 38th International Performance Computing and Communications Conference, IPCCC 2019 [8958720] (2019 IEEE 38th International Performance Computing and Communications Conference, IPCCC 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IPCCC47392.2019.8958720