Concurrent Number Cruncher : An Efficient Sparse Linear Solver on the GPU

Luc Buatois 1 Guillaume Caumon 2 Bruno Lévy 1
1 ALICE - Geometry and Lighting
INRIA Lorraine, LORIA - Laboratoire Lorrain de Recherche en Informatique et ses Applications
2 CRPG - Centre de Recherches Pétrographiques et Géochimiques
Abstract : A wide class of geometry processing and PDE resolution methods needs to solve a linear system, where the non-zero pattern of the matrix is dictated by the connectivity matrix of the mesh. The advent of GPUs with their ever-growing amount of parallel horsepower makes them a tempting resource for such numerical computations. This can be helped by new APIs (CTM from ATI and CUDA from NVIDIA) which give a direct access to the multithreaded computational resources and associated memory bandwidth of GPUs; CUDA even provides a BLAS implementation but only for dense matrices (CuBLAS). However, existing GPU linear solvers are restricted to specific types of matrices, or use non-optimal compressed row storage strategies. By combining recent GPU programming techniques with supercomputing strategies (namely block compressed row storage and register blocking), we implement a sparse generalpurpose linear solver which outperforms leading-edge CPU counterparts (MKL / ACML).
Type de document :
Communication dans un congrès
Ronald Perrott, Barbara M. Chapman, Jaspal Subhlok, Rodrigo Fernandes de Mello, Laurence T. Yang. High Performance Computation Conference - HPCC'07, Sep 2007, Houston, United States. Springer Berlin / Heidelberg, 4782, pp.358-371, 2007, Lecture Notes in Computer Science. 〈10.1007/978-3-540-75444-2_37〉
Liste complète des métadonnées

Littérature citée [17 références]  Voir  Masquer  Télécharger
https://hal.inria.fr/inria-00186833
Contributeur : Nicolas Ray <>
Soumis le : lundi 12 novembre 2007 - 15:58:59
Dernière modification le : mercredi 25 juillet 2018 - 09:59:34
Document(s) archivé(s) le : lundi 12 avril 2010 - 01:57:26

Fichier

HPCC_number_cruncher.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

Collections

INRIA | UNIV-LORRAINE | LORIA | OTELO-UL | CRPG-UL

Citation

Luc Buatois, Guillaume Caumon, Bruno Lévy. Concurrent Number Cruncher : An Efficient Sparse Linear Solver on the GPU. Ronald Perrott, Barbara M. Chapman, Jaspal Subhlok, Rodrigo Fernandes de Mello, Laurence T. Yang. High Performance Computation Conference - HPCC'07, Sep 2007, Houston, United States. Springer Berlin / Heidelberg, 4782, pp.358-371, 2007, Lecture Notes in Computer Science. 〈10.1007/978-3-540-75444-2_37〉. 〈inria-00186833〉

Exporter

BibTeX TEI DC DCterms EndNote

Partager

Métriques

Consultations de la notice

366

Téléchargements de fichiers

326

Contact


archive-ouverte@inria.fr