A Class of Communication-avoiding Algorithms for Solving General Dense Linear Systems on CPU/GPU Parallel Machines

We study several solvers for the solution of general linear systems where the main objective is to reduce the communication overhead due to pivoting. We first describe two existing algorithms for the LU factorization on hybrid CPU/GPU architectures. The first one is based on partial pivoting and the...

Full description

Saved in:

Bibliographic Details
Published in:	Procedia computer science Vol. 9; pp. 17 - 26
Main Authors:	Baboulin, Marc, Donfack, Simplice, Dongarra, Jack, Grigori, Laura, Rémy, Adrien, Tomov, Stanimire
Format:	Journal Article
Language:	English
Published:	Elsevier B.V 2012
Subjects:	communication-avoiding algorithms dense linear algebra libraries hybrid multicore/GPU computing linear system solvers LU factorization linear system solvers dense linear algebra libraries LU factorization communication-avoiding algorithms hybrid multicore/GPU computing
ISSN:	1877-0509, 1877-0509
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	We study several solvers for the solution of general linear systems where the main objective is to reduce the communication overhead due to pivoting. We first describe two existing algorithms for the LU factorization on hybrid CPU/GPU architectures. The first one is based on partial pivoting and the second uses a random preconditioning of the original matrix to avoid pivoting. Then we introduce a solver where the panel factorization is performed using a communication-avoiding pivoting heuristic while the update of the trailing submatrix is performed by the GPU. We provide performance comparisons and tests on accuracy for these solvers on current hybrid multicore-GPU parallel machines.
ISSN:	1877-0509 1877-0509
DOI:	10.1016/j.procs.2012.04.003