FAVERGE, Mathieu; LACOSTE, Xavier; RAMET, Pierre

Métadonnées

Licence d’utilisation du document

FAVERGE, Mathieu
Innovative Computing Laboratory [Knoxville] [ICL]

LACOSTE, Xavier
Parallel tools for Numerical Algorithms and Resolution of essentially Hyperbolic problems [BACCHUS]

RAMET, Pierre
Parallel tools for Numerical Algorithms and Resolution of essentially Hyperbolic problems [BACCHUS]
Laboratoire Bordelais de Recherche en Informatique [LaBRI]

Langue

Rapport

Ce document a été publié dans

2010-05-10p. 22

Résumé en anglais

Over the past few years, parallel sparse direct solvers made significant progress and are now able to solve efficiently industrial three-dimensional problems with several millions of unknowns. To solve efficiently these problems, PaStiX and WSMP solvers for example, provide an hybrid MPI-thread implementation well suited for SMP nodes or multi-core architectures. It enables to drastically reduce the memory overhead of the factorization and improve the scalability of the algorithms. However, today's modern architectures introduce new hierarchical memory accesses that are not handle in these solvers. We present in this paper three improvements on PaStiX solver to improve the performance on modern architectures : memory allocation, communication overlap and dynamic scheduling and some results on numerical test cases will be presented to prove the efficiency of the approach on NUMA architectures.< Réduire

Mots clés en anglais

sparse direct solver

NUMA architecture

multi-cores

dynamic scheduling

URI

https://oskar-bordeaux.fr/handle/20.500.12278/198150

Project ANR

Adaptation et Optimisation des Performances Applicatives sur architectures NUMA. Etude et Mise en Œuvre sur des Applications en SISmologie. - ANR-05-CIGC-0002

Origine

Importé de hal

Unités de recherche

Laboratoire Bordelais de Recherche en Informatique (LaBRI) - UMR 5800