Skip to main content
SpringerLink
Log in

On the benefits and limitations of dynamic partitioning in parallel computer systems

  • Conference paper
  • First Online:
Book cover Job Scheduling Strategies for Parallel Processing (JSSPP 1995)

Part of the book series: Lecture Notes in Computer Science ((LNCS,volume 949))

Included in the following conference series:

Abstract

In this paper we analyze the benefits and limitations of dynamic partitioning across a wide range of parallel system environments. We formulate a general model of dynamic partitioning that can be fitted to measurement data to obtain a sufficiently accurate quantitative analysis of real parallel systems executing real scientific and/or commercial workloads. An exact solution of the model is obtained by employing matrix-geometric techniques. We then use this framework to explore the parallel system design space over which dynamic partitioning outperforms other space-sharing policies for a diverse set of application work-loads, quantifying the significant performance improvements within these regions. Our results show that these regions and the performance benefits of dynamic partitioning are heavily dependent upon its associated costs, the system load, and the workload characteristics. We also identify the regions of the design space over which dynamic partitioning performs poorly, quantifying the performance degradation and illustrating forms of unstable thrashing.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Log in via an institution

Institutional subscriptions

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. S. Asmussen, O. Nerman, and M. Olsson. Fitting phase type distributions via the EM algorithm. Tech. Rep. 1994:23, Dept. Math., Chalmers Univ. Tech., 1994.

    Google Scholar 

  2. S. L. Brumelle. Some inequalities for parallel-server queues. Op. Res., 19:402–413, 1971.

    Google Scholar 

  3. S.-H. Chiang, R. K. Mansharamani, and M. K. Vernon. Use of application characteristics and limited preemption for run-to-completion parallel processor scheduling policies. In Proc. ACM SIGMETRICS Conf., 33–44, 1994.

    Google Scholar 

  4. K. Dussa, B. Carlson, L. Dowdy, and K.-H. Park. Dynamic partitioning in transputer environments. In Proc. ACM SIGMETRICS Conf., 203–213, 1990.

    Google Scholar 

  5. M. J. Faddy. Fitting structured phase-type distributions. Tech. Rep., Dept. Math., Univ. Queensland, Australia, 1994. To appear, Appl. Stock. Mod. Data Anal.

    Google Scholar 

  6. D. G. Feitelson and B. Nitzberg. Job characteristics of a production parallel scientific workload on the NASA Ames iPSC/860. In Job Scheduling Strategies for Parallel Processing, D. G. Feitelson and L. Rudolph (eds.), Springer-Verlag, 1995. Lecture Notes in Computer Science Vol. 949.

    Google Scholar 

  7. G. C. Fox, M. A. Johnson, G. A. Lyzenga, S. W. Otto, J. K. Salmon, and D. W. Walker. Solving Problems on Concurrent Processors Volume I: General Techniques and Regular Problems. Prentice Hall, 1988.

    Google Scholar 

  8. D. Ghosal, G. Serazzi, and S. K. Tripathi. The processor working set and its use in scheduling multiprocessor systems. IEEE Trans. Soft. Eng., 17:443–453, 1991.

    Google Scholar 

  9. A. Gupta, A. Tucker, and S. Urushibara. The impact of operating system scheduling policies and synchronization methods on the performance of parallel applications. In Proc. ACM SIGMETRICS Conf., 1991.

    Google Scholar 

  10. L. Kleinrock. Queueing Systems Volume I: Theory. John Wiley and Sons, 1975.

    Google Scholar 

  11. L. Kleinrock. Queueing Systems Volume II: Computer Applications. John Wiley and Sons, 1976.

    Google Scholar 

  12. A. Lang. Parameter estimation for phase-type distributions, part I: Fundamentals and existing methods. Tech. Rep. 159, Dept. Stats., Oregon State Univ., 1994.

    Google Scholar 

  13. A. Lang and J. L. Arthur. Parameter estimation for phase-type distributions, part II: Computational evaluation. Tech. Rep. 160, Dept. Stats., Oregon State Univ., 1994.

    Google Scholar 

  14. S. T. Leutenegger and M. K. Vernon. The performance of multiprogrammed multi-processor scheduling policies. In Proc. ACM SIGMETRICS Conf., 226–236, 1990.

    Google Scholar 

  15. R. K. Mansharamani and M. K. Vernon. Properties of the EQS parallel processor allocation policy. Tech. Rep. 1192, Univ. Wisconsin, Comp. Sci. Dept., 1993.

    Google Scholar 

  16. C. McCann, R. Vaswani, and J. Zahorjan. A dynamic processor allocation policy for multiprogrammed shared-memory multiprocessors. ACM Trans. Comp. Syst., 11(2):146–178, 1993.

    Google Scholar 

  17. C. McCann and J. Zahorjan. Processor allocation policies for message-passing parallel computers. In Proc. ACM SIGMETRICS Conf., 19–32, 1994.

    Google Scholar 

  18. N. H. Naik, V. K. Naik, and M. Nicoules. Parallelization of a class of implicit finite difference schemes in computational fluid dynamics. Intl. J. High-Speed Comp., 5, 1993.

    Google Scholar 

  19. V. K. Naik. Performance effects of load imbalance in parallel CFD applications. In Proc. SIAM Conf. Par. Proc., 1992.

    Google Scholar 

  20. V. K. Naik. Scalability issues for a class of CFD applications. In Proc. Scal. High Perf. Comp. Conf., 268–275, 1992.

    Google Scholar 

  21. V. K. Naik, S. K. Setia, and M. S. Squillante. Performance analysis of job scheduling policies in parallel supercomputing environments. In Proc. Supercomputing '93, 824–833, 1993.

    Google Scholar 

  22. V. K. Naik, S. K. Setia, and M. S. Squillante. Scheduling of large scientific applications on distributed memory multiprocessor systems. In Proc. SIAM Conf. Par. Proc. Sci. Comp., 913–922, 1993.

    Google Scholar 

  23. R. D. Nelson and M. S. Squillante. The MAtrix-Geometric qUeueing model Solution package (MAGUS) user manual. Tech. Rep. RC, IBM Res. Div., 1994.

    Google Scholar 

  24. M. F. Neuts. Matrix-Geometric Solutions in Stochastic Models: An Algorithmic Approach. The Johns Hopkins Univ. Press, 1981.

    Google Scholar 

  25. E. Rosti, E. Smirni, L. W. Dowdy, G. Serazzi, and B. M. Carlson. Robust partitioning policies of multiprocessor systems. Perf. Eval., 19:141–165, 1994.

    Google Scholar 

  26. R. Schassberger. Insensitivity of steady-state distributions of generalized semi-Markov processes, part I. Ann. Prob., 5(1):87–99, 1977.

    Google Scholar 

  27. R. Schassberger. Insensitivity of steady-state distributions of generalized semi-Markov processes, part II. Ann. Prob., 6(1):85–93, 1978.

    Google Scholar 

  28. R. Schassberger. Insensitivity of steady-state distributions of generalized semi-Markov process with speeds. Advs. Appl. Prob., 10:836–851, 1978.

    Google Scholar 

  29. R. Schreiber and H. D. Simon. Towards the teraflops capability for CFD. In H. D. Simon, editor, Parallel CFD — Implementations and Results Using Parallel Computers. MIT Press, 1992.

    Google Scholar 

  30. S. K. Setia. Scheduling on Multiprogrammed, Distributed Memory Parallel Computers. PhD thesis, Dept. Comp. Sci., Univ. Maryland, College Park, MD, 1993.

    Google Scholar 

  31. S. K. Setia, M. S. Squillante, and S. K. Tripathi. Processor scheduling on multiprogrammed, distributed memory parallel computers. In Proc. ACM SIGMETRICS Conf., 158–170, 1993.

    Google Scholar 

  32. S. K. Setia, M. S. Squillante, and S. K. Tripathi. Analysis of processor allocation in multiprogrammed, distributed-memory parallel processing systems. IEEE Trans. Par. Dist. Syst., 5(4):401–420, 1994.

    Google Scholar 

  33. S. K. Setia and S. K. Tripathi. A comparative analysis of static processor partitioning policies for parallel computers. In Proc. MASCOTS '93, 1993.

    Google Scholar 

  34. K. C. Sevcik. Characterizations of parallelism in applications and their use in scheduling. In Proc. ACM SIGMETRICS Conf., 171–180, 1989.

    Google Scholar 

  35. K. C. Sevcik. Application scheduling and processor allocation in multiprogrammed parallel processing systems. Perf. Eval., 19:107–140, 1994.

    Google Scholar 

  36. M. S. Squillante. MAGIC: A computer performance modeling tool based on matrix-geometric techniques. In Proc. Intl. Conf. Mod. Tech. Tools Comp. Perf. Eval., 411–425, 1991.

    Google Scholar 

  37. M. S. Squillante. Analysis of dynamic partitioning in parallel systems. Tech. Rep. RC 19950, IBM Res. Div., 1995.

    Google Scholar 

  38. M. S. Squillante. On the benefits and limitations of dynamic partitioning in parallel computer systems. Tech. Rep. RC 19951, IBM Res. Div., 1995.

    Google Scholar 

  39. C. A. Thekkath and H. M. Levy. Limits to low-latency communication on high-speed networks. ACM Trans. Comp. Syst., 11(2):179–203, 1993.

    Google Scholar 

  40. A. Tucker and A. Gupta. Process control and scheduling issues for multiprogrammed shared-memory multiprocessors. In Proc. ACM Symp. Op. Syst. Prin., 159–166, 1989.

    Google Scholar 

  41. T. von Eicken, D. E. Culler, S. C. Goldstein, and K. E. Schauser. Active messages: A mechanism for integrated communication and computation. In Proc. Intl. Symp. Comp. Arch., 256–266, 1992.

    Google Scholar 

  42. J. Zahorjan and C. McCann. Processor scheduling in shared memory multiprocessors. In Proc. ACM SIGMETRICS Conf., 214–225, 1990.

    Google Scholar 

Download references

Author information

Authors and Affiliations

  1. IBM T. J. Watson Research Center, 10598, Yorktown Heights, NY, USA

    Mark S. Squillante

Authors
  1. Mark S. Squillante
    View author publications

    You can also search for this author in PubMed Google Scholar

Editor information

Dror G. Feitelson Larry Rudolph

Rights and permissions

Reprints and permissions

Copyright information

© 1995 Springer-Verlag Berlin Heidelberg

About this paper

Cite this paper

Squillante, M.S. (1995). On the benefits and limitations of dynamic partitioning in parallel computer systems. In: Feitelson, D.G., Rudolph, L. (eds) Job Scheduling Strategies for Parallel Processing. JSSPP 1995. Lecture Notes in Computer Science, vol 949. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-60153-8_31

Download citation

  • DOI: https://doi.org/10.1007/3-540-60153-8_31

  • Published:

  • Publisher Name: Springer, Berlin, Heidelberg

  • Print ISBN: 978-3-540-60153-1

  • Online ISBN: 978-3-540-49459-1

  • eBook Packages: Springer Book Archive

Publish with us

Policies and ethics

Search

Navigation