Modeling 1D distributed-memory dense kernels for an asynchronous multifrontal sparse solver

Abstract : To solve sparse systems of linear equations, multifrontal methods rely on dense partial LU decompositions of so-called frontal matrices; we consider a parallel asynchronous setting in which several frontal matrices can be factored simultaneously. In this context, to address performance and scalability issues of acyclic pipelined asynchronous factorization kernels, we study models to revisit properties of left and right-looking variants of partial \(LU\) decompositions, study the use of several levels of blocking, before focusing on communication issues. The general purpose sparse solver MUMPS has been modified to implement the proposed algorithms and confirm the properties demonstrated by the models.
Type de document :
Communication dans un congrès
High Performance Computing for Computational Science, 2014 - 11th International Conference, Revised Selected Papers, Jun 2014, Eugene, Oregon, United States. pp.156-169, 2High Performance Computing for Computational Science, 2014 - 11th International Conference, Eugene, Oregon, USA, June 30 - July 3, 2014, Revised Selected Papers
Liste complète des métadonnées

https://hal.inria.fr/hal-01355356
Contributeur : Equipe Roma <>
Soumis le : mardi 23 août 2016 - 09:52:24
Dernière modification le : jeudi 8 février 2018 - 11:10:04

Identifiants

  • HAL Id : hal-01355356, version 1

Citation

Patrick R. Amestoy, Jean-Yves L'Excellent, François-Henry Rouet, Wissam M. Sid-Lakhdar. Modeling 1D distributed-memory dense kernels for an asynchronous multifrontal sparse solver. High Performance Computing for Computational Science, 2014 - 11th International Conference, Revised Selected Papers, Jun 2014, Eugene, Oregon, United States. pp.156-169, 2High Performance Computing for Computational Science, 2014 - 11th International Conference, Eugene, Oregon, USA, June 30 - July 3, 2014, Revised Selected Papers. 〈hal-01355356〉

Partager

Métriques

Consultations de la notice

292