Adaptive Methods for Hyperbolic Problems on Local Memory Parallel Processors

  • William Gropp
Conference paper
Part of the The IMA Volumes in Mathematics and Its Applications book series (IMA, volume 13)


Parallel computers have opened up new directions for algorithm development. But the variety of parallel architectures and programming methods has lead to two apparently different models of computation. In this paper, I show that these models can be considered as special cases of memory access, similar to data cacheing. The numerical solution of hyperbolic partial differential equations is uses as an example, since simple algorithms for these problems exist and are representative of more realistic algorithms.


Parallel Algorithm Shared Memory Message Passing Mesh Point Critical Section 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    D. J. Baxter and W. D. Gropp, Some Pipelined algorithms for the solution of dense linear systems on hypercubes, 1987. in preparation.Google Scholar
  2. [2]
    Kai Li, Shared Virtual Memory on Loosely Coupled Multiprocessors, Technical Report YALE/DCS/RR-492, Yale University, Department of Computer Science, September 1986.Google Scholar
  3. [3]
    C. L. Seitz, The Cosmic Cube, Communications of the ACM, 28/1 (1985), pp. 22–33.MathSciNetCrossRefGoogle Scholar

Copyright information

© Springer-Verlag New York Inc. 1988

Authors and Affiliations

  • William Gropp
    • 1
  1. 1.Yale UniversityUSA

Personalised recommendations