[gmx-users] g_tune_pme crashes in 5.0.1
mark.j.abraham at gmail.com
Mon Sep 15 22:44:18 CEST 2014
Addressing this problem started back in May, but is stalled in code review
waiting for people who care to finish reviewing it. The basic problem is
that the optimizer would like to check that mdrun works on the .tpr, so it
can tell you that up front before the optimization starts. But there's no
reliable way to choose mdrun -npme for that test, since -1 is likely to
fail at low parallelism (prime factors too common, as here) and 0 is likely
to fail at high parallelism (domains are too small). gmx tune_pme should
grow some more brains, but that depends on my code getting reviewed :-)
The "good" news is that you can hack line 1388 of
src/gromacs/gmxana/gmx_tune_pme.c so that
gmx_bool bFirst = TRUE;
gmx_bool bFirst = FALSE;
and the subsequent compile and install of gmx tune_pme won't do this check
(and is otherwise normal).
On Mon, Sep 15, 2014 at 3:54 PM, Oliver Schillinger <
o.schillinger at fz-juelich.de> wrote:
> I was using g_tune_pme successfully with the 4.6.* branch.
> Now I switched to 5.0.1 and g_tune_pme keeps crashing.
> perf.out (default filenames used) says:
> Cannot run the benchmark simulations! Please check the error message of
> mdrun for the source of the problem. Did you provide a command line
> argument that neither g_tune_pme nor mdrun understands? Offending command:
> mpiexec -np 24 mdrun_mpi -npme 7 -s 04_prod01_bench00.tpr -resetstep 100
> -o bench.trr -x bench.xtc -cpo bench.cpt -c bench.gro -e bench.edr -g
> bench.log -nsteps 1 -quiet
> bench.log contains only output of the first benchmark run with the error
> Fatal error:
> The number of ranks you selected (17) contains a large prime factor 17.
> The problem seems to me that the second run does not get started at all,
> but the program terminates after the first unsuccessful benchmark.
> Am I alone with this problem?
> Oliver Schillinger
> PhD student
> ICS-6 - Structural Biochemistry
> Building 5.8v, Room 3010
> Phone: +49 2461-61-9532
> Mobile: +49 172 53 27 914
> Forschungszentrum Juelich GmbH
> 52425 Juelich
> Sitz der Gesellschaft: Juelich
> Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498
> Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher
> Geschaeftsfuehrung: Prof. Dr. Achim Bachem (Vorsitzender),
> Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt,
> Prof. Dr. Sebastian M. Schmidt
> 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