[gmx-users] md simulation working fine in serial and badly in parallel

Ester Chiessi ester.chiessi at uniroma2.it
Mon Jan 9 10:46:59 CET 2012


 Dear Gromacs People

I am working about md simulations of a linear polymer chain in water at high hydration degree. The force field is G45a3, SPC water, NPT ensemble, gromacs 4.5.3.

I acquired a 50 ns trajecory ( time step 2 fs, all bonds constrained with LINCS, PME, T=293 K, P= 1 atm ) with absolutely no problem using gromacs in serial on my PC.

Therefore I was very surprised when, trying to repeat exactly the same simulation on another computer where mdrun works in parallel (with MPI), the molecular dynamics calculation could not start.

Even working on only 2 nodes, the run crashes at the first step, returning a long list of LINCS errors and an infinite system energy.

Firstly I verified that the processed topology after grompp was the same on the two machines.

Then I tried the parallel run using particle decomposition instead of domain decomposition, without success.

The starting configuration was energy minimized to a maximum force of 90 kJmol-1nm-1. Is this value too high for a parallel simulation?

I can provide information about the mdp file and the mdrun errors of my parallel simulation, in case.

Thanks in advance for any suggestion.

Ester



Ester Chiessi
Dipartimento di Scienze e Tecnologie Chimiche
Università di Roma Tor Vergata
Via della Ricerca Scientifica
00133 Roma
http://www.stc.uniroma2.it/cfmacro/cfmacroindex.htm
tel: *39*6*72594462
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20120109/cafe1b40/attachment.html>


More information about the gromacs.org_gmx-users mailing list