Français Anglais
Accueil Annuaire Plan du site
Accueil > Evenements > Séminaires
Séminaire du LRI
On the Challenges of GPU Computing for Sparse Linear Solvers
Karl Rupp

18 December 2012, 10h30 - 18 December 2012, 12h00
Salle/Bat : 455/PCRI-N
Contact : marc.baboulin@lri.fr

Activités de recherche :

Résumé :
While GPUs are well-established accelerators for dense linear
algebra, they have only been partially successful for the solution of linear systems represented by sparse matrices. An overview of GPU implementations of iterative solvers is given first, then the
challenges of efficient GPU-based preconditioners are discussed.
Subtle differences of the GPU computing frameworks CUDA and OpenCL are presented and library-aspects are introduced by the example of the free open-source library ViennaCL [1] and its current integration into PETSc [2].

[1] http://viennacl.sourceforge.net/
[2] http://www.mcs.anl.gov/petsc/

Pour en savoir plus :
Séminaires
Resilient PDE solving approaches for exascale comp
Calcul à haute performance
Tuesday 29 May 2018 - 10h30
Salle : 465 - PCRI-N
Paul Mycek .............................................

Binary pattern of length greater than 14 are abeli
Combinatoire
Friday 25 May 2018 - 14h30
Salle : 445 - PCRI-N
Matthieu Rosenfeld .............................................

TBA
Algorithmique distribuée
Wednesday 02 May 2018 - 10h30
Salle : 465 - PCRI-N
Evangelos Bampas .............................................

Mariage stable auto-stabilisant et distribué
Théorie des graphes
Friday 13 April 2018 - 14h30
Salle : 445 - PCRI-N
Marie Laveau .............................................

Modélisation et implémentation du produit de matri
Calcul à haute performance
Wednesday 11 April 2018 - 10h30
Salle : 465 - PCRI-N
Thomas Lambert .............................................