A simple performance model for multithreaded applications executing on non-uniform memory access computers

R. Yang, J. Antony, A. P. Rendell

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

11 Citations (Scopus)

Abstract

In this work, we extend and evaluate a simple performance model to account for NUMA and bandwidth effects for single and multi-threaded calculations within the Gaussian 03 computational chemistry code on a contemporary multi-core, NUMA platform. By using the thread and memory placement APIs in Solaris, we present results for a set of calculations from which we analyze on-chip interconnect and intra-core bandwidth contention and show the importance of load-balancing between threads. The extended model predicts single threaded performance to within 1% errors and most multi-threaded experiments within 15% errors. Our results and modeling shows that accounting for bandwidth constraints within user-space code is beneficial.

Original languageEnglish
Title of host publication2009 11th IEEE International Conference on High Performance Computing and Communications, HPCC 2009
Pages79-86
Number of pages8
DOIs
Publication statusPublished - 20 Nov 2009
Externally publishedYes
Event11th IEEE International Conference on High Performance Computing and Communications, HPCC 2009 - Seoul, Korea, Republic of
Duration: 25 Jun 200927 Jun 2009

Publication series

Name2009 11th IEEE International Conference on High Performance Computing and Communications, HPCC 2009

Conference

Conference11th IEEE International Conference on High Performance Computing and Communications, HPCC 2009
CountryKorea, Republic of
CitySeoul
Period25/06/0927/06/09

Keywords

  • performance modelling
  • NUMA
  • multithreaded application

Fingerprint Dive into the research topics of 'A simple performance model for multithreaded applications executing on non-uniform memory access computers'. Together they form a unique fingerprint.

Cite this