Advances in the Solution of NS Equations in GPU Hardware

Abstract

Graphic Processing Units have received much attention in last years. Compute-intensive algorithms operating on multidimensional arrays that have nearest neighbor dependency and/or exploit data locality can achieve massive speedups. An incompressible flow solver has been presented by the authors including a Fractional Step Method, with a Fast Fourier Transform solver for the Poisson stage. In this presentation we describe some improvements to the algorithm, including a parallel implementation of the Fast Marching algorithm for free surface flows.