Toward performance models of MPI implementations for understanding application scaling issues

Torsten Hoefler, William Gropp, Rajeev Thakur, Jesper Larsson Träff

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

Abstract

Designing and tuning parallel applications with MPI, particularly at large scale, requires understanding the performance implications of different choices of algorithms and implementation options. Which algorithm is better depends in part on the performance of the different possible communication approaches, which in turn can depend on both the system hardware and the MPI implementation. In the absence of detailed performance models for different MPI implementations, application developers often must select methods and tune codes without the means to realistically estimate the achievable performance and rationally defend their choices. In this paper, we advocate the construction of more useful performance models that take into account limitations on network-injection rates and effective bisection bandwidth. Since collective communication plays a crucial role in enabling scalability, we also provide analytical models for scalability of collective communication algorithms, such as broadcast, allreduce, and all-to-all. We apply these models to an IBM Blue Gene/P system and compare the analytical performance estimates with experimentally measured values.

Original languageEnglish (US)
Title of host publicationRecent Advances in the Message Passing Interface - 17th European MPI Users' Group Meeting, EuroMPI 2010, Proceedings
Pages21-30
Number of pages10
DOIs
StatePublished - 2010
Event17th European MPI Users' Group Meeting, EuroMPI 2010 - Stuttgart, Germany
Duration: Sep 12 2010Sep 15 2010

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6305 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other17th European MPI Users' Group Meeting, EuroMPI 2010
Country/TerritoryGermany
CityStuttgart
Period9/12/109/15/10

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Toward performance models of MPI implementations for understanding application scaling issues'. Together they form a unique fingerprint.

Cite this