Integrating GPUs as fast co-processors into the existing parallel FE package FEAST
Universitat Dortmund
19th Symposium Simulation Technique (ASIM’06), Hannover, Germany, Sep. 2006
@inproceedings{Goeddeke:2006:IGA,
author={Dominik G{“o}ddeke and Christian Becker and Stefan Turek},
title={Integrating {GPUs} as fast co-processors into the parallel {FE} package {FEAST}},
booktitle={19th Symposium Simulationstechnique (ASIM’06)},
year={2006},
series={Frontiers in Simulation},
editor={Matthias Becker and Helena Szczerbicka},
pages={277–282},
month={sep}
}
We report on our experiences with integrating GPUs as fast, parallel floating-point coprocessors into the parallel FE package FEAST. Since a full re-implementation of such a package is not feasible, we identify the smoothing of an outer domain-decomposition multigrid solver as a natural entry-point for a minimally invasive integration of GPUs. We address the issue of limited computational precision with a mixed precision iterative refinement approach and present preliminary timing results on a commodity cluster enhanced with GPUs.
March 2, 2011 by hgpu
