Automatically tuned collective communications

Sathish S. Vadhiyar, Graham E. Fagg, Jack Dongarra

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

97 Scopus citations

Abstract

The performance of the MPI's collective communications is critical in most MPI-based applications. A general algorithm for a given collective communication operation may not give good performance on all systems due to the differences in architectures, network parameters and the storage capacity of the underlying MPI implementation. In this paper, we discuss an approach in which the collective communications are tuned for a given system by conducting a series of experiments on the system. We also discuss a dynamic topology method that uses the tuned static topology shape, but re-orders the logical addresses to compensate for changing run time variations. A series of experiments were conducted comparing our tuned collective communication operations to various native vendor MPI implementations. The use of the tuned collective communications resulted in about 30%-650% improvement in performance over the native MPI implelementations.

Original languageEnglish
Title of host publicationSC 2000 - Proceedings of the 2000 ACM/IEEE Conference on Supercomputing
PublisherAssociation for Computing Machinery
ISBN (Electronic)0780398025
DOIs
StatePublished - 2000
Externally publishedYes
Event2000 ACM/IEEE Conference on Supercomputing, SC 2000 - Dallas, United States
Duration: Nov 4 2000Nov 10 2000

Publication series

NameProceedings of the International Conference on Supercomputing
Volume2000-November

Conference

Conference2000 ACM/IEEE Conference on Supercomputing, SC 2000
Country/TerritoryUnited States
CityDallas
Period11/4/0011/10/00

Funding

This work was supported by the US Department of Energy through contract number DE-FG02-99ER25378.

FundersFunder number
U.S. Department of EnergyDE-FG02-99ER25378

    Fingerprint

    Dive into the research topics of 'Automatically tuned collective communications'. Together they form a unique fingerprint.

    Cite this