8661

Communication-Avoiding Optimization of Geometric Multigrid on GPUs

Amik Singh
EECS Department, University of California, Berkeley
University of California, Technical Report No. UCB/EECS-2012-258, 2012

@mastersthesis{Singh:EECS-2012-258,

   Author={Singh, Amik},

   Editor={Demmel, James},

   Title={Communication-Avoiding Optimization of Geometric Multigrid on GPUs},

   School={EECS Department, University of California, Berkeley},

   Year={2012},

   Month={Dec},

   URL={http://www.eecs.berkeley.edu/Pubs/TechRpts/2012/EECS-2012-258.html},

   Number={UCB/EECS-2012-258}

}

Download Download (PDF)   View View   Source Source   

739

views

Multigrid methods are widely used to accelerate the convergence of iterative solvers for linear systems in a number of different application areas. In this report, we explore communication-avoiding implementations of Geometric Multigrid on Nvidia GPUs. We achieved an overall gain of 1.2x for the whole multigrid algorithm over baseline implementation. We also provide an insight into what future GPUs need to have in terms of on chip and shared memory for these kinds of algorithms to perform even better.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2017 hgpu.org

All rights belong to the respective authors

Contact us: