A class of communication-avoiding algorithms for solving general dense linear systems on CPU/GPU parallel machines - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Reports (Research Report) Year : 2012

A class of communication-avoiding algorithms for solving general dense linear systems on CPU/GPU parallel machines

Abstract

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 for these solvers on current hybrid multicore-GPU parallel machines.
Fichier principal
Vignette du fichier
RR-7854.pdf (502.94 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-00656457 , version 1 (04-01-2012)
hal-00656457 , version 2 (07-01-2012)
hal-00656457 , version 3 (29-02-2012)

Identifiers

  • HAL Id : hal-00656457 , version 3

Cite

Marc Baboulin, Simplice Donfack, Jack Dongarra, Laura Grigori, Adrien Rémy, et al.. A class of communication-avoiding algorithms for solving general dense linear systems on CPU/GPU parallel machines. [Research Report] RR-7854, INRIA. 2012. ⟨hal-00656457v3⟩
355 View
378 Download

Share

Gmail Facebook X LinkedIn More