[gmx-users] high load imbalance in GMX5.1 using MARTINI FF

Szilárd Páll pall.szilard at gmail.com
Thu Aug 9 13:58:16 CEST 2018


This should indeed normally not happen, but before diving deep into the
issue I'd suggest testing more recent releases of GROMACS, preferably
2018.2 so we know if there is an issue in the currently actively supported

Secondly, load imbalance is not necessarily a bad thing if it does not
deteriorate performance over time. Have you verified that the performance
also decreases over time (e.g. if you run with -v you get performance
continuously reported on the console which you can save in/redirect into a


On Thu, Aug 9, 2018 at 10:13 AM Linda Song <wlsonglinda at gmail.com> wrote:

> Hi,
> I am running some coarse grained simulations of a membrane-protein system
> using Martini force field.
> If I ran it using GMX5.1 and the mdp file provided by Martini website for
> GMX5, the load imbalance will gradually increase to over 80% from 4-10% at
> the beginning in a few hours. When I kill the job and restart from where it
> stopped, the load imbalance will go back to 4-10% at the start but increase
> again.
> If I ran the same simulation using GMX4.6 and the mdp file provided by
> Martini website for GMX4.x and the same number cores, such load imbalance
> issue will disappear. The imbalance will remain stable at 4-10%.
> Both GMX4.6 and GMX5.1 I was using were not compiled with GPU
> acceleration. I wondered what gives rise to this load imbalance in GMX5. Is
> it because the domain decomposition is dealt with differently in GMX5 than
> GMX4.6?
> Thanks :)
> Wanling
> --
> Gromacs Users mailing list
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_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-users or
> send a mail to gmx-users-request at gromacs.org.

More information about the gromacs.org_gmx-users mailing list