Data transfers between processes in an SMP system: Performance study and application to MPI

Darius Buntinas, Guillaume Mercier, William Gropp

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

Abstract

This paper focuses on the transfer of large data in SMP systems. Achieving good performance for intranode communication is critical for developing an efficient communication system, especially in the context of SMP clusters. We evaluate the performance of five transfer mechanisms: shared-memory buffers, message queues, the Ptrace system call, kernel module-based copy, and a high-speed network. We evaluate each mechanism based on latency, bandwidth, its impact on application cache usage, and its suitability to support MPI two-sided and one-sided messages.

Original languageEnglish (US)
Title of host publicationICPP 2006
Subtitle of host publicationProceedings of the 2006 International Conference on Parallel Processing
Pages487-494
Number of pages8
DOIs
StatePublished - 2006
Externally publishedYes
EventICPP 2006: 2006 International Conference on Parallel Processing - Columbus, OH, United States
Duration: Aug 14 2006Aug 18 2006

Publication series

NameProceedings of the International Conference on Parallel Processing
ISSN (Print)0190-3918

Other

OtherICPP 2006: 2006 International Conference on Parallel Processing
Country/TerritoryUnited States
CityColumbus, OH
Period8/14/068/18/06

ASJC Scopus subject areas

  • Hardware and Architecture
  • Engineering(all)

Fingerprint

Dive into the research topics of 'Data transfers between processes in an SMP system: Performance study and application to MPI'. Together they form a unique fingerprint.

Cite this