[gmx-users] fatal error with mpi

[Stefano Guglielmo]

Thanks Mark. I must say that I tried reducing timestep (as low as 0.5 fs-1)
and temperature as well (5K both in NvT and NpT) but the simulation crashed
in any case with warning about cutoff, lincs or bad water. The very
previous day I had run the same simulation on the same machine but with the
non-MPI version of gromacs 2016.5 and everything went smoothly; as I
switched to the MPI version I encountered the issue, and in the end I was
afraid to have missed something during compilation, so I decided to try and
recompile and now it is running.
Anyway I will check the trajectory as you said.
Thanks
Stefano

2018-06-22 15:36 GMT+02:00 Mark Abraham <mark.j.abraham at gmail.com>:

> Hi,
>
> Recompiling won't fix anything relevant. Molecular frustration isn't just
> about long bonds. Rearrangment of side chain groupings can do similar
> things despite looking happy. The sudden injection of KE means collisions
> can be more violent than normal, and the timestep is now too large. But you
> need to look at the trajectory to understand if this might be the case.
>
> Mark
>
> On Fri, Jun 22, 2018 at 3:19 PM Stefano Guglielmo <
> stefano.guglielmo at unito.it> wrote:
>
> > Dear Mark,
> >
> > thanks for your answer. The version is 2016.5, but I apparently solved
> the
> > problem recompiling gromacs: now the simulation is running quite stable.
> I
> > had minimized and gradually equilibrated the system and I could not see
> any
> > weird bonds or contacts. So in the end, as extrema ratio, I decided to
> > recompile.
> >
> > Thanks again
> > Stefano
> >
> > 2018-06-22 15:07 GMT+02:00 Mark Abraham <mark.j.abraham at gmail.com>:
> >
> > > Hi,
> > >
> > > This could easily be that your system is actually not yet well
> > equilibrated
> > > (e.g. something was trapped in a high energy state that eventually
> > relaxed
> > > sharply). Or it could be a code bug. What version were you using?
> > >
> > > Mark
> > >
> > > On Thu, Jun 21, 2018 at 2:36 PM Stefano Guglielmo <
> > > stefano.guglielmo at unito.it> wrote:
> > >
> > > > Dear users,
> > > >
> > > > I have installed gromacs with MPI instead of its native tMPI and I am
> > > > encountering the following error:
> > > > "Fatal error:
> > > > 4 particles communicated to PME rank 5 are more than 2/3 times the
> > > cut-off
> > > > out
> > > > of the domain decomposition cell of their charge group in dimension
> y.
> > > > This usually means that your system is not well equilibrated."
> > > >
> > > > I am using 8 MPI ranks with 4 omp threads per rank (it was the
> > > > configuration used on the same machine by tMPI), but I also tried 4
> > ranks
> > > > with 8 threads, but it did not solve the problem.
> > > >
> > > > I don't think this is an issue related to my system because the same
> > > system
> > > > run with the native tMPI works properly (it has already been
> termalized
> > > and
> > > > gradually equilibrated); neverthless I tried to reduce dt and
> > temperature
> > > > but without any benefit. Does anybody have any suggestions?
> > > >
> > > > Thanks in advance
> > > > Stefano
> > > >
> > > > --
> > > > Stefano GUGLIELMO PhD
> > > > Assistant Professor of Medicinal Chemistry
> > > > Department of Drug Science and Technology
> > > > Via P. Giuria 9
> > > > 10125 Turin, ITALY
> > > > ph. +39 (0)11 6707178 <011-670%2071%2078> <011-670%2071%2078>
> > > > --
> > > > 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.
> > > >
> > > --
> > > 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.
> > >
> >
> >
> >
> > --
> > Stefano GUGLIELMO PhD
> > Assistant Professor of Medicinal Chemistry
> > Department of Drug Science and Technology
> > Via P. Giuria 9
> > 10125 Turin, ITALY
> > ph. +39 (0)11 6707178 <011-670%2071%2078>
> > --
> > 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.
> >
> --
> 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.
>



-- 
Stefano GUGLIELMO PhD
Assistant Professor of Medicinal Chemistry
Department of Drug Science and Technology
Via P. Giuria 9
10125 Turin, ITALY
ph. +39 (0)11 6707178