[gmx-users] gromacs 4.5.3 with threads instead of MPI

[Carsten Kutzner]

Hi,

On Jan 17, 2011, at 4:11 PM, Arnau Cordomi wrote:

> Dear Gromacs Users,
> 
> We normally run gromacs 4.0.x mdrun with OpenMPI in a 24 core
> shared-memory server (SunFire X4450).
> i. e. the command we use for a 12 core run is : mpirun -np 12 mdrun -v
> -c output_md.gro
> This is working great so far.
> 
> Now we are trying to use  gromacs 4.5.x and I found this on the
> release notes (http://www.gromacs.org/About_Gromacs/Release_Notes/Versions_4.5.x)
> :
> 
> “Running on a multi-core node now uses thread-based parallelization to
> automatically spawn the optimum number of threads in the default
> build. MPI is now only required for parallelization over the network.”
> 
> So, I guess now instead of “mpirun -np 12 mdrun -v -c output_md.gro”
> we should use “mdrun -nt 12 -v -c output_md.gro” and expect the same
> performance. Am I right?
Right.

> 
> Also, is this “automatically spawn the optimum number of threads”
> reliable ? Does that mean that if the recommended number is 4 cores
> (threads) there’s no way to make it run faster even if specifying -nt
> 12 or 24 ?
You should use the number of logical cores you have. This is normally
the number of physical cores, unless you use some kind of hyperthreading
or simultaneous multiprocessing in which case -nt is twice the number
of physical cores. mdrun reads the number of threads from your system
via sysconf().

Carsten

> 
> Any advice will be welcome.
> Thanks in advance.
> 
> Best Regards,
> Arnau
> 
> Cordomí
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/Support/Mailing_Lists


--
Dr. Carsten Kutzner
Max Planck Institute for Biophysical Chemistry
Theoretical and Computational Biophysics
Am Fassberg 11, 37077 Goettingen, Germany
Tel. +49-551-2012313, Fax: +49-551-2012302
http://www.mpibpc.mpg.de/home/grubmueller/ihp/ckutzne