CFD on graphic cards

Abstract

In the last decade, the entertainment and graphic design industries demand for higher resolution and more realistic graphics has motivated graphic card manufacturers to develop high performance graphic processing units (GPUs) at low cost. Nowadays GPUs are highly efficient parallel processing units that can be used for general purposes. Such developments open a new alternative within scientific computation, promising high performance parallel computation at everyones reach. Graphic cards seem to be an attractive option, specially for research groups and institutions with limited computational resources and for teaching purposes. Here, an overview of parallel computation on GPUs is presented, as well as the efforts to manipulate and promote this technology within the UNAM, particularly in Fluid Mechanics applications.