[gmx-developers] repeatability of runs?

[Hardy, Adam]

For me this is on the 5.0.4 release.

Adam Hardy
PhD Student
School of Engineering and Physical Sciences
Heriot-Watt University
Edinburgh EH14 4AS, UK

________________________________________
From: gromacs.org_gmx-developers-bounces at maillist.sys.kth.se [gromacs.org_gmx-developers-bounces at maillist.sys.kth.se] on behalf of Berk Hess [hess at kth.se]
Sent: 16 February 2015 10:50
To: gmx-developers at gromacs.org
Subject: Re: [gmx-developers] repeatability of runs?

Hi,

What version(s) are we talking about here? A released 5.0 or git master
code?
In git there have been some issues with PME with many OpenMP threads and
shortly some GPU issues. But the current git master code should be correct.

The -reprod option tries to remove any source of divergence, such as
FFTW auto-tuning and dynamic load balancing. Any single bit change will
cause differences that diverge exponentially due to the chaotic nature
of MD. But you should not see any "strange" results, such as
pressures/virials that are higher than expected.

Cheers,

Berk

On 2015-02-16 11:37, Hardy, Adam wrote:
> I'm having a similar problem lately. I've been plotting cumulative averages and after a restart (using the full precision binary files) I saw a significant increase in average pressure (>20 bar in a system at ~360 bar). I've made some changes to the virial code which excludes some frozen atoms from the calculation so I had thought it was perhaps an error on my part causing this. I'm also seeing divergence on a similar scale doing multiple simulations from the same .tpr file.
>
> I've done similar stuff before last year on different systems without frozen atoms and not seen this problem, particularly after restarts, I did not see this pressure jump. I'm doing a bit more work today to try and establish whether it's my changes causing this or something else.
>
> I haven't used the -reprod command, should this have such a large effect?
>
> Adam Hardy
> PhD Student
> School of Engineering and Physical Sciences
> Heriot-Watt University
> Edinburgh EH14 4AS, UK
>
> ________________________________________
> From: gromacs.org_gmx-developers-bounces at maillist.sys.kth.se [gromacs.org_gmx-developers-bounces at maillist.sys.kth.se] on behalf of Shirts, Michael R. (mrs5pt) [mrs5pt at eservices.virginia.edu]
> Sent: 16 February 2015 03:31
> To: gmx-developers at gromacs.org
> Subject: [gmx-developers] repeatability of runs?
>
> Hi, all-
>
> Is anyone else having issues with repeatability of runs?  When I start
> from the same tpr using one thread (to avoid load rebalancing) on the same
> machine, I find that it diverges after the first step.  It appears to be
> an issue with the virial.  Any thoughts?  It seems like the code should be
> binary repeatable at this level.  Makes it hard to debug a problem!
>
> Best,
> ~~~~~~~~~~~~
> Michael Shirts
> Associate Professor
> Department of Chemical Engineering
> University of Virginia
> michael.shirts at virginia.edu
> (434) 243-1821
>
> --
> Gromacs Developers mailing list
>
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers or send a mail to gmx-developers-request at gromacs.org.
>
>
> -----
> We invite research leaders and ambitious early career researchers to
> join us in leading and driving research in key inter-disciplinary themes.
> Please see www.hw.ac.uk/researchleaders for further information and how
> to apply.
>
> Heriot-Watt University is a Scottish charity
> registered under charity number SC000278.
>

--
Gromacs Developers mailing list

* Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-developers_List before posting!

* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-developers or send a mail to gmx-developers-request at gromacs.org.


----- 
We invite research leaders and ambitious early career researchers to 
join us in leading and driving research in key inter-disciplinary themes. 
Please see www.hw.ac.uk/researchleaders for further information and how
to apply.

Heriot-Watt University is a Scottish charity
registered under charity number SC000278.