[gmx-users] Re: Gromacs - LAM/MPI

[David]

On Fri, 2002-11-08 at 20:53, Paulo F. B. Goncalves wrote:
> 
> mpirun -v -c 12 -nger -lamd mdrun_mpi -np 12 -s ${MOL}_em.tpr -o
> ${MOL}.trr -c ${MOL}.gro -v -g ${MOL}.log
On normal (fast) ethernet you cannot scale to much more than two nodes
for such small systems. The reason for this is partially the
parallellization in gromacs, but mainly because the force calculation is
so fast relative to the communication. I suggest you use your cluster
for 8 two processor jobs while we work on imporved gromacs scaling.

Please post future questions to the gromacs mailing list.
 

-- 
Groeten, David.
________________________________________________________________________
Dr. David van der Spoel, 	Biomedical center, Dept. of Biochemistry
Husargatan 3, Box 576,  	75123 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://zorn.bmc.uu.se/~spoel
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++