[gmx-users] Error when for gmx tune_pme on Cray

[Mark Abraham]

Hi,

Yes, Cray's MPI also does not permit such nesting. This is why it's
generally advisable that installations make both an MPI and non-MPI build,
and to let the default suffixing of the gmx binary handle itself so both
installations can go to the same place for convenience.

Mark

On Tue, Apr 10, 2018 at 3:50 PM Kutzner, Carsten <ckutzne at gwdg.de> wrote:

> Hi Viveca,
>
> > On 10. Apr 2018, at 15:10, Viveca Lindahl <vivecalindahl at gmail.com>
> wrote:
> >
> > Hi users,
> >
> > I never used gmx tune_pme before and thought I'd try. On a Cray machine,
> > using aprun instead of mpirun, I did
> >
> > args="/cfs/klemming/nobackup/v/vivecal/programs/gromacs/2018.1/bin/gmx
> > mdrun -pin on -quiet -stepout 5000 -ntomp 2 -noconfout"
> > export MPIRUN=aprun; aprun -n 1 $gmx tune_pme -npstring n -s topol.tpr
> -np
> > 32 -mdrun "$args”
> You don’t need to use aprun -n 1 on $gmx tune_pme, as this is
> just a serial program that itself calls aprun on mdrun_mpi.
> Does the following work:
>
> export MPIRUN=aprun
> $gmx tune_pme -npstring n -s topol.tpr -np 32 -mdrun /path/to/mdrun_mpi
>
> I don’t know if on Cray it is allowed that one MPI executable (tune_pme)
> starts
> another MPI executable (mdrun), IBM PE e.g. does not allow it.
> There, tune_pme must be compiled without MPI, only mdrun can be linked to
> the MPI lib.
>
> Best,
>   Carsten
>
> >
> > It hangs (no output) for some minutes, then I get the error
> >
> > Program:     gmx tune_pme, version 2018.1-dev-20180315-3dfa11852
> > Source file: src/gromacs/gmxana/gmx_tune_pme.cpp (line 740)
> >
> > Fatal error:
> > Cannot execute mdrun. Please check benchtest.log for problems!
> >
> > For more information and tips for troubleshooting, please check the
> GROMACS
> > website at http://www.gromacs.org/Documentation/Errors
> > -------------------------------------------------------
> > Making sure that mdrun can be executed. Trying 'aprun -n 32
> > /cfs/klemming/nobackup/v/vivecal/programs/gromacs/2018.1/bin/gmx mdrun
> -pin
> > on -quiet -stepout 5000 -ntomp 2 -noconfout -version -maxh 0.001 1>
> > benchtest.log 2>&1' ... Rank 0 [Tue Apr 10 14:58:12 2018] [c1-0c0s4n0]
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> >
> > the suggested log file to look in has
> >
> > craylog: WARNING: log tmp dir /var/spool/cray/llm is not writable
> > aprun: Unable to contact apsys
> > aprun: Exiting due to errors. Application aborted
> >
> > Am I doing something wrong or is this a bug/limitation of the system I am
> > trying to run on?
> >
> > Thanks,
> > --
> > Viveca
> > --
> > 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.