[gmx-users] Need help on a SEGV mdrun mpi failure
Mostyn.Lewis at sun.com
Sun Oct 26 02:54:00 CEST 2003
Thanks for the reply. However, it still fails with the latest
CVS (Friday). I thought I'd move this to the developers list.
On Fri, 24 Oct 2003, David wrote:
> On Fri, 2003-10-24 at 18:20, Mostyn Lewis wrote:
> > Hello,
> > Sent this last night but it seems to have been lost (Maybe because it
> > had a 270K attachment of topol.top.bz2?). So here goes again.
> > I'm having a problem with a benchmark case which causes SEGV (signal 11)
> > in most cases of a MPI run with more than 4 CPUs. The failure is always
> > in bondfree.c (gromacs-3.1.4 + gromacs-3.1.5_pre1) in the angles routine
> > at line 535
> > ivec_sub(SHIFT_IVEC(g,ai),jt,dt_ij);
> > ivec_sub(SHIFT_IVEC(g,ak),jt,dt_kj);
> > t1=IVEC2IS(dt_ij);
> > t2=IVEC2IS(dt_kj);
> > rvec_inc(fr->fshift[t1],f_i);
> > rvec_inc(fr->fshift[CENTRAL],f_j);
> > ----->rvec_inc(fr->fshift[t2],f_k);
> > } /* 168 TOTAL */
> > This line has a BAD t2 value which causes an out of bounds reference
> > (actually a little later in x=a[XX]+b[XX]; at line 235 in vec.h due to
> > the expansion of rvec_inc)
> thanks for the bug report... Had I had this half a year ago, then it
> would have saved me a lot of time. As it is, this has been fixed in the
> CVS version of the code (unless I'm terribly wrong), as I assume that
> this is a run with a protein in little water...
> > I enclose a run below with the grompp and mdrun_mpi output followed by
> > some dbx debugging output showing some values. This was on a 24 CPU
> > SUN SMP box (Sunfire 6800) using 8 CPUs.
> > I get the same failure on a cluster of Linux (2 CPU Xeon) boxes doing
> > MPI across Gigabit ethernet. The failure occurs in Linux land using
> > Intel/PGI and LAM/mpich combinations - so I think this is problem and/or
> > Gromacs dependent.
> > I'm not a Molecular persona at all, just a humble benchmarker and seek
> > help from the enlightened.
> > Any files you'd like (topol.top ...) or more debugging are available
> > on request.
> > Sorry this is so long. Any help would be appreciated.
> David van der Spoel, PhD, Assist. Prof., Molecular Biophysics group,
> Dept. of Cell and Molecular Biology, Uppsala University.
> Husargatan 3, Box 596, 75124 Uppsala, Sweden
> phone: 46 18 471 4205 fax: 46 18 511 755
> spoel at xray.bmc.uu.se spoel at gromacs.org http://xray.bmc.uu.se/~spoel
> gmx-users mailing list
> gmx-users at gromacs.org
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
More information about the gromacs.org_gmx-users