[gmx-users] MPI issue

[Carsten Kutzner]

Hi Paul,

is this version 3.3.1? If I am not missing something, the only two
instances where MPI_Allreduce could be used are in pme.c and in
network.c, but the respective parts of the code have to be explicitly
switched on by a preprocessor directive. Anyway, the communicator on
which they operate is MPI_COMM_WORLD and this should never be NULL.

Or does the error happen at the very end of the simulations? Then maybe
it's a problem with MPI_Finalize (you might want to have a look at
gmx_finalize in network.c then).

If you are using the CVS version things are different, because Allreduce
is used more often. Is there additional information in the error report
that tells in which routine the problem occurs?

Carsten



Paul Whitford wrote:
> I am running gromacs on a machine with altivec support.  My
> understanding is I have to use MVAPICH on this machine.
> 
> I occasionally have runs crash and I receive the following error
> 
> 
> n559(0): Loaded with Money
> 
> n536(14): 14 - MPI_ALLREDUCE : Null communicator
> n536(14): [14] [] Aborting Program!
> n559(0): 0 - MPI_ALLREDUCE : Null communicator
> n559(0): [0] [] Aborting Program!
> Cleaning up all processes ...Cleaning up all processes ...done.
> 
> 
> Does this indicate a problem with the machine, mvapich, or gromacs? 
> Again, this doesn't always happen, just sometimes (which makes me
> suspect it is a machine issue).  Any advice on where the problem might
> be would be helpful.  thanks in advance.
> 
> -Paul
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/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/mailing_lists/users.php

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