[gmx-users] REMD restart: writing 2nd traj part on same local disk

[David van der Spoel]

pascal.baillod at epfl.ch wrote:
> Dear community,
> 
> I am running gmx3.3 REMD on a 32 node cluster, writing data to the local node
> disks, with the following command:
> 
> mpiexec -np 64 $MDRUN -multi -np 64 -s $INPUT/md.tpr -o $OUT/traj.trr -x
> $OUT/traj.xtc -c $OUT/conf.g96 -e $OUT/ener.edr -g $OUT/md.log -replex 5000
> -reseed -1 >& log-job &
> 
> ...where $OUT is the path to a directory on the local node hard disk, and $INPUT
> the path to a /home directory accessible by all nodes and containing all the
> .tpr files. The node local hard disks are not cross-mounted by NFS for sake of
> performance, and are therefore only visible when logged uppon the corresponding
> node. 
> 
> Following a crash, I restarted the simulation by preparing restart .tpr files.
> For every temperature X, my script sends the gromacs tpbconv command to node X,
> where the trr trajectory file of temperature X is stored and can be read, and
> writes the restart .tpr file for temperature X somewhere on /home. I can then
> restart the simulation using these .tpr with a command line similar to the one
> printed above.
> 
> The problem is that the temperature-trajectories no longer are processed by the
> same node as before. In other words, after the restart, node X runs temperature
> Y MD instead of running temperature X MD. I wonder if that can be corrected, so
> as to retrieve the original order (with node X running the 2nd part of
> temperature X MD, as it already ran the first part) and facilitate the pasting
> of the 1st and 2nd parts of the simulation.
> 
> Thanks for any hint!!
> 
> Pascal
> 
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> 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
The name of the node is printed in each md*.log file. Depending on your 
MPI implementation it may or may not be possible to supply the list (and 
order) of nodes that you want for your continuation run.

Since you have written the log file on the local disk too, you will need 
a smart script to generate new tprs.

-- 
David.
________________________________________________________________________
David van der Spoel, PhD, Assoc. Prof., Molecular Biophysics group,
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,  	75124 Uppsala, Sweden
phone:	46 18 471 4205		fax: 46 18 511 755
spoel at xray.bmc.uu.se	spoel at gromacs.org   http://folding.bmc.uu.se
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++