[gmx-users] Problem using MPIRUN MDRUN

Mark Abraham Mark.Abraham at anu.edu.au
Mon Feb 16 11:26:41 CET 2009

fprisch at nimr.mrc.ac.uk wrote:
> I did not stop the simulation manually and it did not give a segmentation fault message, but something else that I have never seen before (and actually, I am still not understanding the error message).
> Now I'm running the dynamics on my machine, instead of running it on the cluster, and there are no problems at all. This means that the problem is not related to my protein or to the parameters of the dynamic, but it must be somewhere else in the mpirun command line or something like that.

OK, so there's probably some problem with file system availability on 
the cluster, and/or returning output files to the user. We can't help 
you there.

This sort of detail and differential diagnosis would have been a good 
thing to say the first time you described your problem. See the final 
link here - http://wiki.gromacs.org/index.php/Support

> grompp_mpi_d -f md.mdp -c file_pr.gro -p file.top -o file.tpr -np 32
> mpirun -np 32 /dms/prog/bin/mdrun_mpi_s -s file.tpr -o file.trr -c file.gro  -np 32

This looks like asking for trouble, if _s and _d are your suffixes for 
single and double. Don't mix them. If you've erroneously retyped them, 
then that's not helpful information for us. Computers are literal, and 
we can only help you work out what you've told it wrongly if you tell us 
the same thing - so write a script and/or copy-paste things so that you 
are reproducible.


More information about the gromacs.org_gmx-users mailing list