Abstract
With increasing number of processors the visualization of trace data for MPI communication primitives becomes harder due to many limitations. An automatic analysis of the recorded data should be able to identify some clues for the optimization of the source code. The approach presented with this paper is able to divide the communication time for each communication relationship into a necessary part and the overhead. We use statistical methods to define a time for each single communication event that we will accept as the ‘usual’ time for this event. Additional runtime will be considered as overhead. At the end of this article, the benefits from this method are demonstrated on the Sweep3D benchmark.
Chapter PDF
Similar content being viewed by others
References
Ahn, D.H., Vetter, J.S.: Scalable analysis techniques for microprocessor performance counter metrics. In: Proceedings of Supercomputing 2002, pp. 1–16 (2002)
OpenMP Architecture Review Board. OpenMP Application Program Interface, Version 2.5 (November 2004), http://www.openmp.org/drupal/mp-documents/draft_spec25.pdf
DeRose, L., Wolf, F.: CATCH - A Call-Graph Based Automatic Tool for Capture of Hardware Perfromance Metrics for MPI and OpenMP Applications. In: Monien, B., Feldmann, R.L. (eds.) Euro-Par 2002. LNCS, vol. 2400. Springer, Heidelberg (2002)
Message Passing Interface Forum. MPI: A Message-Passing Interface Standard. Technical report, University of Tennessee (1995)
Juckeland, G., Börner, S., Kluge, M., Kölling, S., Nagel, W.E., Pflüger, S., Rödling, H., Seidl, S., William, T., Wloch, R.: BenchIt-Performance Measurement and Comparison for Scientific Applications. In: Joubert, G.R., et al. (eds.) Advances in Parallel Computing, vol. 13, pp. 501–508. Elsevier, Amsterdam (2004)
Kluge, M.: Statistische Analyse von Programmspuren für MPI-Programme. Diploma thesis (November 2004)
Knüpfer, A.: A New Data Compression Technique for Event Based Program Traces. In: International Conference on Computational Science 2003, pp. 956–965 (2003)
Knüpfer, A., Nagel, W.E.: Compressible Memory Data Structures for Event Based Trace Analysis. In: Future Generation Computer Systems. Elsevier, Amsterdam (2004)
Lawrence Livermore National Laboratory. The ASCI Sweep3D Benchmark Code (1995), http://www.llnl.gov/asci_benchmarks/asci/limited/sweep3d/asci_sweep3d.html
: Tom LeBlanc and Wagner Meira. Measurement and Prediction of Parallel Program Performance (1997) http://www.cs.rochester.edu/u/leblanc/prediction.html , http://www.cs.rochester.edu/u/leblanc/prediction.html
Nagel, W.E., Arnold, A., Weber, M., Hoppe, H.-C., Solchenbach, K.: VAMPIR: Visualization and Analysis of MPI Resources. In: Supercomputer 63, vol. XII(1), pp. 69–80 (1996)
Sachs, L.: Applied Statistics: A Handbook of Techniques, 11th edn. Springer, Heidelberg (2002)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2005 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Kluge, M., Knüpfer, A., Nagel, W.E. (2005). Statistical Methods for Automatic Performance Bottleneck Detection in MPI Based Programs. In: Sunderam, V.S., van Albada, G.D., Sloot, P.M.A., Dongarra, J.J. (eds) Computational Science – ICCS 2005. ICCS 2005. Lecture Notes in Computer Science, vol 3514. Springer, Berlin, Heidelberg. https://doi.org/10.1007/11428831_41
Download citation
DOI: https://doi.org/10.1007/11428831_41
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-26032-5
Online ISBN: 978-3-540-32111-8
eBook Packages: Computer ScienceComputer Science (R0)