Is GPU the future of Scientific Computing?

Georges-Henri Cottet (1) , Jean-Matthieu Etancelin (1) , Franck Pérignon (2) , Christophe Picard (1) , Florian de Vuyst (3) , Christophe Labourdette (3)

1 EDP - Equations aux Dérivées Partielles
2 LJK - Laboratoire Jean Kuntzmann
3 CMLA - Centre de Mathématiques et de Leurs Applications

In this article we present a graphics processing unit (GPU) implementation of a particle method for transport equations. More precisely the numerical method under consideration is a remeshed particle method. Not only remeshing particles makes simulations more accurate in flows with strong strain, but it leads to algorithms more regular in term of data structures. In this work, we develop a Python library using GPU through OpenCL standard that implements this remeshed particle method which already shows interesting performances.

Mots clés: GPU; particle method; advection equation; python; opencl

Domaines: Analyse numérique [math.NA]

Georges-Henri Cottet, Jean-Matthieu Etancelin, Franck Pérignon, Christophe Picard, Florian de Vuyst, et al.. Is GPU the future of Scientific Computing?. Annales Mathématiques Blaise Pascal, 2013, 20 (1), pp.75-99. ⟨10.5802/ambp.322⟩. ⟨hal-00780565⟩