Advertisement

A Proposal of Reconfigurable MPI Collective Communication Functions

Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 3758)

Abstract

Message Passing Interface (MPI) Collective Communication Functions (MCCF) are usually implemented in programming libraries utilizing invariable algorithms. Not always do such algorithms yield the best performance with all kinds of applications and over all execution environments. In this paper, we present, simulate, analytically model, verify and analyze reconfigurable MCCF that present variable structures and behaviors, in order to provide optimized configurations, flexibility and performance. In this paper we propose and present a set of optimized reconfigurable MCCF, which add flexibility and high performance to collective communications. We simulate, analytically model, verify and analyze the proposed functions, and compare them with invariable implementations. Our results show that reconfiguration at the algorithm level really yields flexibility and performance gains in MCCF.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Gorlatch, S.: Send-Receive Considered Harmful: Myths and Realities of Message Passing. In: ACM TOPLAS 2004, vol. 26(1), pp. 47–56 (2004) ISSN:0164-0925Google Scholar
  2. 2.
    Huse, L.P.: Collective Communication on Dedicated Clusters of Workstations. In: Euro PVM/MPI, pp. 469–476 (1999)Google Scholar
  3. 3.
    Tan, W.B., Strazdins, P.: The Analysis and Optimization of Collective Communications on a Beowulf Cluster. In: ICPADS, pp. 659–666 (2002)Google Scholar
  4. 4.
    Thakur, R., Gropp, W.: Improving the Performance of Collective Operations in MPICH. In: Euro PVM/MPI, pp. 257–267 (2003)Google Scholar
  5. 5.
    Vadhiyar, S.S., Fagg, G.E., Dongarra, J.: Automatically Tuned Collective Communications. In: SuperComputing 2000: HPNC (2000) CD-ROM, ISBN 0-7803-9802-5Google Scholar
  6. 6.
    Squyres, J.M., Lumsdaine, A.: A Component Architecture for LAM/MPI. In: Euro PVM/MPI, pp. 379–387 (2003)Google Scholar
  7. 7.
    Kielmann, T., et al.: Network performance-aware collective communication for clustered wide-area systems. Parallel Computing 27(11), 1431–1456 (2001)zbMATHCrossRefGoogle Scholar
  8. 8.
    Barnett, M., et al.: Building a High-Performance Collective Communication Library. SuperComputing, 107–116 (1994)Google Scholar
  9. 9.
    Ramos, L.E.S., Martins, C.A.P.S.: Reconfigurable MPI Collective Communication Functions. In: WSCAD 2004 (2004) (in Portuguese)Google Scholar
  10. 10.
    Freitas, H.C.: Proposal and Development of a Network Processor with Reconfigurable Crossbar Switch, M.Sc. thesis, PPGEE, PUC-MG (2003) (in Portuguese)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2005

Authors and Affiliations

  1. 1.Rutgers University 
  2. 2.Pontifical Catholic University of Minas Gerais 

Personalised recommendations