Supporting dynamic data and processor repartitioning for irregular applications

  • José E. Moreira
  • Kalluri Eswar
  • Ravi B. Konuru
  • Vijay K. Naik
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1117)


Recent research has shown that dynamic reconfiguration of resources allocated to parallel applications can improve both system utilization and application throughput. Distributed Resource Management System (DRMS) is a parallel programming environment that supports development and execution of reconfigurable applications on a dynamically varying set of resources. This paper describes DRMS support for developing reconfigurable irregular applications, using a sparse Cholesky factorization as a model application. We present performance levels achieved by DRMS redistribution primitives, which show that the cost of dynamic data redistribution between different processor configurations for irregular data are comparable to those for regular data.


Irregular applications dynamic data distribution reconfigurable partitions DRMS 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Agerwala, T., Martin, J. L., Mirza, J. H., Sadler, D. C., Dias, D. M., and Snir, M., SP2 system architecture. IBM Systems Journal, 34(2): 152–184, 1995.Google Scholar
  2. 2.
    Chapman, B., Mehrotra, P., and Zima, H. User defined mappings in Vienna Fortran. SIG-PLAN Notices, 28(1), 1993.Google Scholar
  3. 3.
    Duff, I. S., Grimes, R. G., and Lewis, J. G., Sparse matrix test problems. ACM Transactions on Mathematical Software, 15:1–14, 1989.CrossRefGoogle Scholar
  4. 4.
    Edjlali, E., Agrawal, G., Sussman, A., and Saltz, J. Data parallel programming in an adaptive environment. In Proceedings of 9th International Parallel Processing Symposium, Santa Barbara, CA, April 1995.Google Scholar
  5. 5.
    Fox, G., Hiranandani, S., Kennedy,K., Koelbel, C., Kremer, U., Tseng, C., and Wu, M. Fortran D language specification. Technical Report COMP TR90-141, Department of Computer Science, Rice University, December 1990.Google Scholar
  6. 6.
    Indiana University. Sage++, A Class library for Building Fortran 90 and C++ Restructuring Tools, May 1995.Google Scholar
  7. 7.
    Koelbel, C. H., Loveman, D. B., Schreiber, R. S., Steele Jr., G. L., and Zosel, M. E. The High Performance Fortran Handbook. The MIT Press, 1994.Google Scholar
  8. 8.
    Kohn, S. R., and Baden, S. B. A robust parallel programming model for dynamic non-uniform scientific computations. In Proceedings of Scalable High-Performance Computing Conference, pages 509–517, Knoxville, TN, May 1994.Google Scholar
  9. 9.
    Konuru, R., Casas, J., Otto, S. W., Prouty, R., and Walpole, J. A user-level process package for PVM. In Proceedings of the Scalable High Performance Computing Conference, pages 48–55, Knoxville, TN, May 1994.Google Scholar
  10. 10.
    Konuru, R. B., Moreira, J. E., and Naik, V. K. Application-assisted dynamic scheduling on large-scale multi-computer systems. Technical Report RC 20390, IBM Research Division, February 1996. To appear in Euro-Par'96, Lyon, France, August 27–29, 1996.Google Scholar
  11. 11.
    Lain, A. and Banerjee, P. Exploiting spatial regularity in irregular iterative applications. In Proceedings of the 9th International Parallel Processing Symposium, pages 820–827, Santa Barbara, CA, April 1995.Google Scholar
  12. 12.
    McCann, C., Vaswami, R., and Zahorjan, J. A dynamic processor allocation policy for multiprogrammed shared-memory multiprocessors. ACM Transactions on Computer Systems, 11(2):146–178, May 1993.CrossRefGoogle Scholar
  13. 13.
    Müller, A. and Rühl, R. Extending High Performance Fortran for the support of unstructured computations. In Proceedings of the 1995 International Conference on Supercomputing, pages 127–136, July 3–7 1995.Google Scholar
  14. 14.
    Naik, V. K., Setia, S. K., and Squillante, M. S. Processor allocation in multiprogrammed, distributed-memory parallel computer systems. Technical Report RC 20239, IBM Research Division, October 1995. Submitted to Journal of Parallel and Distributed Computing.Google Scholar
  15. 15.
    Ramaswamy, S. and Banerjee, P. Processor allocation and scheduling of macro dataflow graphs on distributed memory multicomputers by the PARADIGM compiler. In Proceedings of the International Conference on Parallel Processing, pages II:134–138, August 1993.Google Scholar
  16. 16.
    Sadayappan, P. and Visvanathan, V. Distributed sparse factorization of circuit matrices via recursive e-tree partitioning. In SIAM Symposium on Sparse Matrices, Gleneden Beach, OR, 1989.Google Scholar
  17. 17.
    Sharma, S. D., Ponnusamy, R., Moon, B., Hwang, Y., Das, R., and Saltz, J. Run-time and compile-time support for adaptive irregular problems. In Proceedings of Supercomputing'94, pages 97–106, November 1994.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1996

Authors and Affiliations

  • José E. Moreira
    • 1
  • Kalluri Eswar
    • 1
  • Ravi B. Konuru
    • 1
  • Vijay K. Naik
    • 1
  1. 1.IBM T. J. Watson Research CenterYorktown Heights

Personalised recommendations