[gmx-developers] FW: repeatability of runs?

Berk Hess hess at kth.se
Mon Feb 16 20:48:35 CET 2015

Note that with the VV integrator issue 1669 was fixed recently in the 
5.0 branch, but that should only affect MPI parallel runs of small systems.



On 02/16/2015 07:53 PM, Shirts, Michael R. (mrs5pt) wrote:
>> Note that we don't guarantee that -reprod will give reproducable runs.
>> But without a GPU (and excluding the possibility of non-deterministic
>> MPI reductions), I don't see anything that could interfere with
>> reproducibility.
> Yes, that's what I would expect.  Note that I don't think that we need
> MANY ways that the trajectories should be reproducible -- but it would be
> useful if there was one way that it was.
>> Note that restarting from checkpoint files should
>> result in reproducible results. This can help with debugging, since you
>> don't need to run all those ns, only from the last checkpoint and you
>> can then reduce the checkpoint interval until it crashes in a few steps.
> I agree, they should, but they haven't been, at least with this example,
> on my machine (I'll add some more details in the issue).  I've started off
> with just the tpr, running on one thread, with the assumption that that's
> is simpler than .cpt.
> Best,
> ~~~~~~~~~~~~
> Michael Shirts
> Associate Professor
> Department of Chemical Engineering
> University of Virginia
> michael.shirts at virginia.edu
> (434) 243-1821

More information about the gromacs.org_gmx-developers mailing list