Abstract
A common statement in papers in the vectorization field is to note that point SOR methods with the natural ordering cannot be vectorized. The usual approach is to reorder the unknowns using a red-black or diagonal ordering and vectorize that. In this paper we construct a point Gauss-Seidel iteration which completely vectorizes and still uses the natural ordering. The work here also applies to both point SOR and single program, multiple data (SPMD) parallel computer architectures. When this approach is reasonable to use is also shown.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
L. Hayes, Comparative analysis of iterative techniques for solving LaPlace’s equation on the unit square on a parallel processor, Master’s Thesis, University of Texas, Austin, TX (1974).
J.J. Lambiotte and R.G. Voigt, The solution of tridiagonal linear systems on the CDC STAR-100 computer, ACM Trans. Math. Soft. 1(1975)308–329.
R.A. Sweet, A parallel and vector variant of the cyclic reduction algorithm, SIAM J. Sci. Stat. Comp. 9(1988)761–765.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Douglas, C.C. Some remarks on completely vectorizing point Gauss-Seidel while using the natural ordering. Adv Comput Math 2, 215–222 (1994). https://doi.org/10.1007/BF02521108
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF02521108