Solution of alternating-line processes on modern parallel computers

Author

Espadas, David ; Prieto, Manuel ; Llorente, Ignacio M. ; Tirado, Francisco

Author_Institution

Fac. de Ciencias Fisicas, Univ. Complutense de Madrid, Spain

fYear

1999

fDate

1999

Firstpage

208

Lastpage

215

Abstract

The aim of this paper is the study of different methods for the solution of alternating-line problems, taking into account the evolution of architectural parameters on modern parallel computers, i.e. processors, memory hierarchy, and interconnection network performance. Three different kinds of solvers are studied: The Pipelined Gaussian Elimination scheme, the Matrix Transposition scheme, and the new one which is presented in this paper: the Mapping Transposition scheme, whose performance clearly betters, in many cases, that obtained by all the other methods, due to its better fitting to the characteristics of modern parallel computers. The experimental results have been obtained on a Cray T3E and on an SGI Origin 2000, up to 512 and 32 processors, respectively

Keywords

parallel processing; performance evaluation; Cray T3E; Mapping Transposition scheme; Matrix Transposition scheme; Pipelined Gaussian Elimination scheme; SGI Origin 2000; alternating-line processes; architectural parameters; interconnection network performance; memory hierarchy; parallel computers; Anisotropic magnetoresistance; Concurrent computing; Difference equations; Differential equations; Ear; Finite difference methods; Iterative methods; Multiprocessor interconnection networks; Partial differential equations; Smoothing methods;

fLanguage

English

Publisher

ieee

Conference_Titel

Parallel Processing, 1999. Proceedings. 1999 International Conference on

Conference_Location

Aizu-Wakamatsu City

ISSN

0190-3918

Print_ISBN

0-7695-0350-0

Type

conf

DOI

10.1109/ICPP.1999.797406

Filename

797406