[MMTK] mpi minimization

Konrad Hinsen hinsen at cnrs-orleans.fr
Tue Oct 21 15:01:07 UTC 2008


On Oct 21, 2008, at 13:47, Lars Skjærven wrote:

> I want to perform an all-atom normal mode calculation with MMTK.  
> For that reason I need to perform a minimization which takes some  
> time on a single CPU.

Do you have a multiprocessor (or multicore) machine? In that case,  
use multiple threads first. That's simpler than MPI and gives you  
much better speedups.

MPI parallelization in MMTK has never been very well optimized. We  
are working on a completely new system, but that won't be ready  
rapidly. Multithreading works pretty well though, I get a speedup of  
3.8 on a four-core machine.

Konrad.
--
---------------------------------------------------------------------
Konrad Hinsen
Centre de Biophysique Moléculaire, CNRS Orléans
Synchrotron Soleil - Division Expériences
Saint Aubin - BP 48
91192 Gif sur Yvette Cedex, France
Tel. +33-1 69 35 97 15
E-Mail: hinsen at cnrs-orleans.fr
Web: http://dirac.cnrs-orleans.fr/~hinsen/
---------------------------------------------------------------------






More information about the mmtk mailing list