[gmx-users] mpich - mdrun problem and md using mpich

[Yuhua Song]

Dear Colleague:
 
When I try to run Gromacs 3.2.1 on supercomputer with multiprocessor,
After I launch the simulation using 6 processors or more it doesn't
work. The odd thing is that it doesn't exit with errors, but it waits
for data which never arrive. Finally report the error about Timeout
ect.
 
/opt/mpich/gnu/bin/mpirun -np 6 -nolocal -machinefile machines
/usr/local/bin/mdrun_mpi -np 6 -v
and the output is something like
 
starting mdrun 'Generated by genbox'
100 steps,      0.2 ps.
and, after some time, it writes out
p4_1636:  p4_error: Timeout in establishing connection to remote
process:
 
 
I search the mailing list, the same issue that has been reported on
mailing list on 2004-10-18 by gianluca santarossa titled as: "md using
mpich".
 
David van der Spoel 's reply indicate that "there is a problem in mpich
that makes that jobs never end, it has to do with interaction between
MPI and grid software. There is a workaround in the CVS code that will
shortly be released in beta." I am wondering whether current public
version of 3.2.1 have solved this problem, the error comes from the MPI
installation.  Or need to download special patch for this.
 
Thank you very much,
 
 
Yuhua
 
 
 
 
 
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20050810/02538e2b/attachment.html>