Analyzing the parallel program efficiency based on monitoring data

Authors

  • D.A. Nikitenko
  • K.S. Stefanov

Keywords:

high-performance computing
supercomputers
efficiency of applications and supercomputers
monitoring
performance analysis

Abstract

An approach to the analysis of parallel program efficiency is proposed on the basis of monitoring data available on Linux-based OS running parallel program. Some criteria of low efficiency and its classification are given. The results of a number of experiments are discussed.

New computational technologies

Downloads

Published

2012-09-10

Issue

Vol. 13 (2012): Issue 3.

Section

Section 2. Programming

Author Biographies

D.A. Nikitenko

Lomonosov Moscow State University,
Research Computing Center
• Junior Researcher

K.S. Stefanov

Lomonosov Moscow State University,
Research Computing Center
• Researcher

References

  1. Walden J., Messer A., Kuhl A. Idea: Measuring the Effect of Code Complexity on Static Analysis Results // Proc. of ESSoS. Vol. 5429. Berlin: Springer, 2009. 195-199.
  2. How to calculate IOPS for your storage system? (http://www.techshell.org/2010/04/11/calculate_iops/).