[gmx-users] REMD and mpi

pascal.baillod at epfl.ch pascal.baillod at epfl.ch
Tue Dec 27 22:34:00 CET 2005

Dear community,

I am currently testing the gromacs-3.3 implementation of REMD and have the
following questions..

1) Is it possible to write the different REMD trajectories on different
locations? I would be interested in writing the trajectories locally, that is,
on the disk next to the compute node, on a machine that dose not have fast
switches. If that is not possible, is there any benchmark on the bottleneck
created by many compute nodes writing REMD data on a same disk, on a machine
that dose not have fast switches (max 1gbps)?

2) I have tried running NPT-REMD with 3 replicas, at 300, 305 and 310 K. The
barostat of my 3 .mdp input files has rep_p = 1.0. I get the following error
Fatal error:
The reference pressure decreases with increasing temperature

If I do the same test with the first 2 replicas only (300 and 305K), I do not
get this error message and things seem to run fine. 

3) Is there any script to get statistics of swaps? In order to compute the
temperature series visited by a replica, for example..

Thank you very much for your help!


Pascal Baillod (PhD student) 
Swiss Federal Institute of Technology EPFL	        Tel: +41-(0)21-693-0322
Institute of Chemical Sciences and Engineering ,	Fax: +41-(0)21-693-0320
Laboratory of Computational Chemistry and Biochemistry	pascal.baillod at epfl.ch
Room BCH 4121, Avenue Forel,	                        http://lcbcpc21.epfl.ch
CH-1015 Lausanne	

More information about the gromacs.org_gmx-users mailing list