Elsevier

Parallel Computing

Volume 21, Issue 3, March 1995, Pages 401-422
Parallel Computing

Paper
The cost of eigenvalue computation on distributed-memory MIMD multiprocessors☆

https://doi.org/10.1016/0167-8191(94)00100-OGet rights and content

Abstract

In [20], Simon proves that bisection is not the optimal method for computing an eigenvalue on a single vector processor. In this paper, we show that his analysis does not extend in a straightforward way to the computation of an eigenvalue on a distributed-memory MIMD multiprocessor. In particular, we show how the optimal number of sections (and processors) to use for multisection depends on variables such as the matrix size and the ratio of communication and computation costs. We also show that parallel multisection outperforms the variant of parallel bisection called polysection proposed by Swarztrauber in [22] for this problem on a distributed-memory MIMD multiprocessor. We present the results of experiments on the 64-processor Intel iPSC/2 hypercube and the 512-processor Intel Touchstone Delta mesh multiprocessor.

References (22)

  • J. Dongarra et al.

    Block reduction of matrices to condensed form for eigenvalue computations

    J. Comp. Appl. Math.

    (1989)
  • M. Avriel
  • M. Barnett et al.

    Optimal broadcasting in mesh-connected architectures

  • W. Barth et al.

    Calculation of the eigenvalues of a symmetric tridiagonal matrix by the method of bisection

  • H. Bernstein et al.

    Optimizing Givens' algorithm for multiprocessors

    SIAM J. Sci. Stat. Comput.

    (1988)
  • J. Dixon

    Estimating extremal eigenvalues and condition numbers of matrices

    SIAM J. Numer. Anal.

    (1983)
  • T. Dunigan

    Performance of the Intel iPSC/860 hypercube

  • W. Givens

    Numerical computation of the characteristic values of a real symmetric matrix

  • H. Huang

    A parallel algorithm for symmetric tridiagonal eigenvalue problems

  • I. Ipsen et al.

    Solving the symmetric tridiagonal eigenvalue problem on the hypercube

    SIAM J. Sci. Stat. Comput.

    (1990)
  • E. Jessup

    Parallel solution of the symmetric tridiagonal eigenproblem

    • Cited by (0)

    ☆

    Both authors were funded by DOE contract DE-FG02-92ER25122 and by NSF grant CCR-9109785. Part of this work was completed when the second author was in residence at Oak Ridge National Laboratory and funded by DOE contract DE-AC05-84OR21400.

    View full text
    Copyright © 1995 Published by Elsevier B.V.