Benchmarking parallel I/O performance for a large scale scientific application on the TeraGrid

Frank Löffler, Jian Tao, Gabrielle Allen, Erik Schnetter

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

Abstract

This paper is a report on experiences in benchmarking I/O performance on leading computational facilities on the NSF TeraGrid network with a large scale scientific application. Instead of focusing only on the raw file I/O bandwidth provided by different machine architectures, the I/O performance and scalability of the computational tools and libraries that are used in current production simulations are tested as a whole, however with focus mostly on bulk transfers. It is seen that the I/O performance of our production code scales very well, but is limited by the I/O system itself at some point. This limitation occurs at a low percentage of the computational size of the machines, which shows that at least for the application used for this paper the I/O system can be an important limiting factor in scaling up to the full size of the machine.

Original languageEnglish (US)
Title of host publicationHigh Performance Computing and Applications - Second International Conference, HPCA 2009, Revised Selected Papers
Pages272-279
Number of pages8
DOIs
StatePublished - 2010
Externally publishedYes
Event2nd International Conference on High-Performance Computing and Applications, HPCA 2009 - Shanghai, China
Duration: Aug 10 2009Aug 12 2009

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5938 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other2nd International Conference on High-Performance Computing and Applications, HPCA 2009
Country/TerritoryChina
CityShanghai
Period8/10/098/12/09

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Benchmarking parallel I/O performance for a large scale scientific application on the TeraGrid'. Together they form a unique fingerprint.

Cite this