Synthetic workload generation for parallel processing systems

  • Hannes Pfneiszl
  • Gabriele Kotsis
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1127)


In the performance evaluation for parallel systems, modeling and generating the workload (i.e. the (set of) programs) is one of the most important and crucial tasks. While benchmarks are frequently used to characterize the real workload in measurement studies, they often fail to adequately describe the real workload, that the analyst has in mind. What is needed is a support for generating synthetic workloads which are on the one hand able to characterize the real workload at the desired level of detail and which are on the other hand easy to construct.

In this paper we describe a tool which has been designed and implemented with respect to these demands. The basic idea is to provide a set of communication patterns (e.g. one-to-one, one-to-all) and computation patterns (“tasks”), which are the building blocks of the synthetic program. By “putting” these blocks together, the analyst can create the desired algorithmic structure. This skeleton is the input to an analysis and simulation tool (N-MAP) [Fers 95a]. Within this environment, various quantitative parameters describing the duration of computations and communications can be specified. The “execution” of the skeleton is then simulated based on the provided parameter values.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [Bail 86]
    D. H. Bailey and J. T. Barton. “The NAS Kernel Benchmark Program”. Tech. Rep., Numerical Aerodynamic Simulation (NAS) Systems Division, NASA Ames Research Center, June 1986.Google Scholar
  2. [Bodn 91]
    R. R. Bodnarchuk and R. B. Bunt. “A Synthetic Workload Model for a Distributed System File Server”. Performance Evaluation Review, Special Issue, 1991 ACM SIGMETRICS, Vol. 19, No. 1, pp. 50–59, May 1991.Google Scholar
  3. [Cand 92]
    R. Candlin, P. Fisk, L. Phillips, and N. Skilling. “Studying the Performance of Concurrent Programs by Simulation Experiments on Synthetic Programs”. Performance Evaluation Review, Special Issue, 1992 ACM SIGMETRICS, Vol. 20, No. 1, pp. 239–241, June 1992. Poster Session of Extended Abstracts.Google Scholar
  4. [Dong 95]
    J. J. Dongarra. “Performance of Various Computers Using Standard Linear Equations Software”. Tech. Rep. CS-89-85, University of Tennessee and Oak Ridge National Laboratory, November 1995.Google Scholar
  5. [Fers 95a]
    A. Ferscha and J. Johnson. “Evaluation of Accuracy/Cost-Tradeoffs in the N-MAP Environment”. Tech. Rep. D3H-4 (GZ 308.926), University of Vienna, PACT Consortium, June 1995.Google Scholar
  6. [Fers 95b]
    A. Ferscha and J. Johnson. “Implementation of Workload Characterization Tools: The N-MAP Environment”. Tech. Rep. D3H-3 (GZ 308.926), University of Vienna, PACT Consortium, June 1995.Google Scholar
  7. [Fuji 90]
    R. M. Fujimoto. “Performance of Time Warp under Synthetic Workloads”. In: D. Nicol, Ed., Distributed Simulation. Proceedings of the SCS Multiconference on Distributed Simulation, pp. 23–28, Society for Computer Simulation, San Diego, California, 1990. Simulation Series, Volume 22, Number 1.Google Scholar
  8. [Geto 93]
    V. S. Getov, A. J. G. Hey, R. W. Hockney, and I. C. Wolton. “The GENESIS Benchmark Suite: Current State and Results”. In: Performance Evaluation of Parallel Systems, PEPS'93, November 29–30, 1993, Warwick, UK, pp. 182–190, University of Warwick, 1993.Google Scholar
  9. [Kao 92]
    W.-I. Kao and R. K. Iyer. “A User-Oriented Synthetic Workload Generator”. In: Proceedings of the 12th International Conference on Distributed Computing Systems, pp. 270–277, IEEE Computer Society Press, Los Alamitos, California, 1992.Google Scholar
  10. [LaRo 92]
    R. P. LaRowe, C. S. Ellis, and M. A. Holiday. “Evaluation of NUMA Memory Management Through Modeling and Measurements”. IEEE Transactions on Parallel and Distributed Systems, Vol. 3, No. 6, pp. 686–701, Nov. 1992.Google Scholar
  11. [Mess 90]
    P. Messina, C. Bailie, P. Hipes, J. Rogers, A. Alagar, A. Kamrath, R. Leary, W. Pfeiffer, R. Williams, and D. Walker. “Benchmarking advanced architecture computers”. Concurrency: Practice and Experience, Vol. 2, No. 3, pp. 195–255, Sep. 1990.Google Scholar
  12. [Pfne 96]
    H. Pfneiszl. Synthetic Workload Generation for Parallel Processing Systems. Master's thesis, University of Vienna, Institute of Applied Computer Science, January 1996.Google Scholar
  13. [Quin 87]
    M. J. Quinn. Designing efficient algorithms for parallel computers. McGraw-Hill International Publishers, New York, 1987.Google Scholar
  14. [Roge 93]
    S. A. Rogers. “A Synthetic Workload Generator for Evaluating Distributed Fault-tolerant Environments”. Tech. Rep. ESL-AFT-040-93, MCC, 1993.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1996

Authors and Affiliations

  • Hannes Pfneiszl
    • 1
  • Gabriele Kotsis
    • 1
  1. 1.Universität WienWienÖsterreich

Personalised recommendations