• DocumentCode
    1668491
  • Title

    The improved Krylov subspace methods for large and sparse linear systems on bulk synchronous parallel architectures

  • Author

    Yang, Laurence Tianruo ; Brent, Richard P.

  • Author_Institution
    Dept. of Comput. Sci., Saint Francis Xavier Univ., Antigonish, NS, Canada
  • fYear
    2003
  • Abstract
    In this paper, we would like to summarize the recent advances on the improved Krylov subspace methods for the solutions of large and sparse linear systems of equations with unsymmetric coefficient matrices. The proposed methods combine elements of numerical stability and parallel algorithm design without increasing much computational costs. The methods have the following common feature that all are derived such that all matrix-vector multiplication, inner products and vector updates of a single iteration step are independent and communication time required for inner product can be overlapped efficiently with computation time of vector updates. Therefore, the cost of global communication which represents the bottleneck of the performance can be significantly reduced. Here, the bulk synchronous parallel (BSP) model is used to design fully efficient, scalable and portable parallel proposed algorithms and to provide accurate performance prediction of the algorithms for a wide range of architectures including the Cray T3D, the Parsytec, and a cluster of workstations connected by an Ethernet. This performance model uses only a few system dependent parameters based on a simple and accurate cost modelling to provide useful insight in the time complexity of the method. The theoretical performance predictions are compared with some preliminary measured timing results of a numerical application from ocean flow simulation.
  • Keywords
    flow simulation; geophysics computing; iterative methods; local area networks; matrix multiplication; numerical stability; oceanographic techniques; parallel algorithms; parallel architectures; parallel programming; performance evaluation; sparse matrices; vectors; workstation clusters; Cray T3D; Ethernet; Parsytec; bulk synchronous parallel architectures; cluster of workstations; improved Krylov subspace methods; inner products; iteration step; large sparse linear systems; matrix-vector multiplication; numerical stability; ocean flow simulation; parallel algorithm design; performance prediction; unsymmetric coefficient matrices; vector updates; Algorithm design and analysis; Clustering algorithms; Costs; Equations; Linear systems; Numerical stability; Parallel algorithms; Parallel architectures; Predictive models; Sparse matrices;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Parallel and Distributed Processing Symposium, 2003. Proceedings. International
  • ISSN
    1530-2075
  • Print_ISBN
    0-7695-1926-1
  • Type

    conf

  • DOI
    10.1109/IPDPS.2003.1213473
  • Filename
    1213473