Design, Tuning and Evaluation of Parallel Multilevel ILU Preconditioners

In this paper, we present a parallel multilevel ILU preconditioner implemented with OpenMP. We employ METIS partitioning algorithms to decompose the computation into concurrent tasks, which are then scheduled to threads. Concretely, we combine decompositions which obtain significantly more tasks tha...

Full description

Saved in:
Bibliographic Details
Published inHigh Performance Computing for Computational Science - VECPAR 2008 pp. 314 - 327
Main Authors Aliaga, José I., Bollhöfer, Matthias, Martín, Alberto F., Quintana-Ortí, Enrique S.
Format Book Chapter Publication
LanguageEnglish
Published Berlin, Heidelberg Springer Berlin Heidelberg 2008
SeriesLecture Notes in Computer Science
Subjects
Incomplete LU factorization
Informàtica
Numerical Algorithms for CS&E
OpenMP
parallel algorithm
Parallel and Distributing Computing
shared-memory multiprocessor
Sparse linear system
Àrees temàtiques de la UPC
Online AccessGet full text
ISBN3540928588
9783540928584
ISSN0302-9743
1611-3349
DOI10.1007/978-3-540-92859-1_28

Cover

More Information
Summary:In this paper, we present a parallel multilevel ILU preconditioner implemented with OpenMP. We employ METIS partitioning algorithms to decompose the computation into concurrent tasks, which are then scheduled to threads. Concretely, we combine decompositions which obtain significantly more tasks than processors, and the use of dynamic scheduling strategies in order to reduce the thread’s idle time, which it is shown to be the main source of overhead in our parallel algorithm. Experimental results on a shared-memory platform consisting of 16 processors report remarkable performance for our approach.
ISBN:3540928588
9783540928584
ISSN:0302-9743
1611-3349
DOI:10.1007/978-3-540-92859-1_28