RZBENCH: Performance Evaluation of Current HPC Architectures Using Low-Level and Application Benchmarks

  • Georg Hager
  • Holger Stengel
  • Thomas Zeiser
  • Gerhard Wellein

Abstract

RZBENCH is a benchmark suite that was specifically developed to reflect the requirements of scientific supercomputer users at the University of Erlangen-Nuremberg (FAU). It comprises a number of application and low-level codes under a common build infrastructure that fosters maintainability and expandability. This paper reviews the structure of the suite and briefly introduces the most relevant benchmarks. In addition, some widely known standard benchmark codes are reviewed in order to emphasize the need for a critical review of often-cited performance results. Benchmark data is presented for the HLRB-II at LRZ Munich and a local InfiniBand Woodcrest cluster as well as two uncommon system architectures: A bandwidth-optimized InfiniBand cluster based on single socket nodes (“Port Townsend”) and an early version of Sun’s highly threaded T2 architecture (“Niagara 2”).

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Standard Performance Evaluation Corporation. http://www.spec.org
  2. 2.
    The NAS parallel benchmarks, published by the NASA Advanced Supercomputing Division. http://www.nas.nasa.gov/Resources/Software/npb.html
  3. 3.
    F. Deserno, G. Hager, F. Brechtefeld, G. Wellein, in High Performance Computing in Science and Engineering Munich 2004, ed. by S. Wagner et al. (Springer, Berlin, 2005), pp. 3–25 CrossRefGoogle Scholar
  4. 4.
  5. 5.
  6. 6.
    Sun Microsystems, OpenSPARC T2 core microarchitecture specification. http://opensparc-t2.sunsource.net/specs/OpenSPARCT2_Core_Micro_Arch.pdf
  7. 7.
    W. Schönauer, Scientific Supercomputing: Architecture and Use of Shared and Distributed Memory Parallel Computers, self-edition (Karlsruhe, 2000). http://www.rz.uni-karlsruhe.de/~rx03/book
  8. 8.
    Sun Microsystems, private communication Google Scholar
  9. 9.
    MPI ping-pong results for HPC challenge. http://icl.cs.utk.edu/hpcc/hpcc_results_lat_band.cgi
  10. 10.
    J. Treibig, G. Hager, Why is performance productivity poor on modern architectures? Talk with Jan Treibig at the Dagstuhl Seminar on Petacomputing, Dagstuhl, Germany, 13–17 February 2006. http://kathrin.dagstuhl.de/files/Materials/06/06071/06071.TreibigJan.Slides.pdf
  11. 11.

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Georg Hager
    • 1
  • Holger Stengel
    • 1
  • Thomas Zeiser
    • 1
  • Gerhard Wellein
    • 1
  1. 1.Regionales Rechenzentrum Erlangen (RRZE)ErlangenGermany

Personalised recommendations