Advertisement

Optimizing OpenMP Parallelized DGEMM Calls on SGI Altix 3700

  • Daniel Hackenberg
  • Robert Schöne
  • Wolfgang E. Nagel
  • Stefan Pflüger
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4128)

Abstract

Using functions of parallelized mathematical libraries is a common way to accelerate numerical applications. Computer architectures with shared memory characteristics support different approaches for the implementation of such libraries, usually OpenMP or MPI.

This paper’s content is based on the performance comparison of DGEMM calls (floating point matrix multiplication, double precision) with different OpenMP parallelized numerical libraries, namely Intel MKL and SGI SCSL, and how they can be optimized. Additionally, we have a look at the memory placement policy and give hints for initializing data. Our attention has been focused on a SGI Altix 3700 Bx2 system using BenchIT [1] as a very convenient performance measurement suite for the examinations.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    BenchIT: Homepage, http://www.benchit.org
  2. 2.
    Juckeland, G., Börner, S., Kluge, M., Kölling, S., Nagel, W.E., Pflüger, S., Röding, H., Seidl, S., William, T., Wloch, R.: ParCo 2003: BenchIT - Performance Measurement and Comparison for Scientic Applications (2003), http://www.benchit.org/DOWNLOAD/DOCUMENTS/parco_paper.pdf
  3. 3.
    Juckeland, G., Kluge, M., Nagel, W.E., Pflüger, S.: Performance Analysis with BenchIT: Portable, Flexible, Easy to Use. In: QEST, pp. S320–321. IEEE Computer Society, Los Alamitos (2004), ISBN 0–7695–2185–1Google Scholar
  4. 4.
    Schöne, R., Juckeland, G., Nagel, W.E., Pflüger, S., Wloch, S.: Parco 2005: Performance comparison and optimization: Case studies using BenchIT (2005), http://www.benchit.org/downloads/documents/parco_05_abstract.pdf
  5. 5.
    Silicon Graphics Inc.: Homepage http://www.sgi.com
  6. 6.
    Oak Ridge National Laboratoy: Evaluation of the Altix 3700 at Oak Ridge National Laboratoy, http://www.gelato.unsw.edu.au/archives/linux-ia64/0409/10993.html
  7. 7.
    University of Tennessee: Basic Linear Algebra Subprograms Technical (BLAST) Forum, http://www.netlib.org/utk/papers/blas-report.ps
  8. 8.
  9. 9.
    Silicon Graphics Inc.: Scientific Computing Software Library, http://www.sgi.com/products/software/scsl.html
  10. 10.
    Silicon Graphics Inc.: Linux Application Tuning Guide, http://techpubs.sgi.com/library/manuals/4000/007-4639-004/pdf/007-4639-004.pdf

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Daniel Hackenberg
    • 1
  • Robert Schöne
    • 1
  • Wolfgang E. Nagel
    • 1
  • Stefan Pflüger
    • 1
  1. 1.Center for Information Services and High Performance Computing (ZIH)Technische Universität DresdenDresdenGermany

Personalised recommendations