[gmx-users] more than 2 step*.pdb files?
feenstra at chem.vu.nl
Wed Jun 4 16:09:01 CEST 2003
Berk Hess wrote:
> A parallel mdrun can indeed write multiple pairs of step*.pdb files.
> I didn't think of this when I implemented it.
> Each processor handles and constrains a different part of the system
> and writes pdb files the constraining of its part of the system does
> not converge.
> It can not be put in one file, since the new coordinates are not
> communicated yet.
> So for parallel runs the cpu number should be added to the file names.
OK, that also fits with my observation that for several of the
files, coordinates are 0. But then how can there be files from three
(or more, for Aldo's system) different time steps? Is that because not
all CPU's decide at the same time that a crash is in progress (i.e.
they each decide for themselves, and may 'disagree')?
| | |
| _ _ ___,| K. Anton Feenstra |
| / \ / \'| | | Dept. of Pharmacochem. - Vrije Universiteit Amsterdam |
|( | )| | | De Boelelaan 1083 - 1081 HV Amsterdam - Netherlands |
| \_/ \_/ | | | Room P275 Tel: +31 20 44 47608 Fax: +31 20 44 47610 |
| | Feenstra at chem.vu.nl - www.chem.vu.nl/~feenstra/ |
| | "Engage" (J.L. Picard) |
More information about the gromacs.org_gmx-users