[gmx-developers] Current status FFTW-3D

Carsten Kutzner ckutzne at gwdg.de
Mon Aug 6 16:50:25 CEST 2007

David van der Spoel wrote:
> Erik Lindahl wrote:
>> Hi Roland,
>> Wiki is pretty much up-to-date on the FFT stuff - I haven't had time
>> for it myself (working on free energy perturbation right now) and my
>> students seem to be busy :-)
>> There are some auxiliary projects involving more optimized FFT
>> routines, but the primary project is mostly an exercise in
>> book-keeping and efficient MPI communication. Essentially, the
>> parallel version should use pre-existing 1D FFT calls and enable full
>> 1D/2D/3D decomposition of the entire grid (currently we only do 1D).
> The wiki is up-to-date as concerns information. Carsten Kutzner
> expressed some interest in working on this, maybe he can comment whether
> or not this is the case?

Hi David, Roland, and Erik,

yes, I expressed interest in working on the FFT parallelization,
however, right now I have started to implement an additional pull
option. I cannot say when exactly I would have time for the FFT - so
please go ahead! I'm shure there are plenty of things to work on so that
I will simply choose something else then.

Thanks for asking!

More information about the gromacs.org_gmx-developers mailing list