Harald, No money, no time at our end. With the absolute latest offerings: double precision and 1+ gigabytes memory this could well be a viable way to significantly higher performance. We plan to start to play around with this in the next few months, outside help would be great. Barry On Sep 17, 2008, at 10:24 AM, Harald Pfeiffer wrote:

Hi,

do you know whether there are any efforts to run PETSc on GPUs (graphical processing units)?

Thanks, Harald

-- Harald P. Pfeiffer [email protected] Theoretical Astrophysics Phone (626) 395-8413 Caltech 130-33 Fax (626) 796-5675 Pasadena, CA 91125, USA