[gmx-users] g_sans_mpi problem: bus error (core dumped)

[Mark Abraham]

On Tue, Mar 24, 2015 at 12:40 PM, CHEN Pan <evan.pan.chen at gmail.com> wrote:

> Hi Mark,
>
> What's the difference between the frames read and the frames analyzed?
> Since g_sans read a frame, it's going to analyze it. Right or no?
>

Depends how badly it's been implemented ;-) I've never looked at that one.
But since the most likely explanation is running out of memory, you can
test for that by observing that something like -b 50000 -e 70000 works but
-e 80000 does not. If -b 0 -e 50000 works, then we know there is a problem
with using memory for frames read but not analyzed. XTC is designed to be
seekable, but there are so many archaic trajectory-reading APIs and
fossilized analysis tools in GROMACS...

Also, running analysis from an MPI build gives you extra ways to have
problems and no advantage, so I recommend not doing that.

Mark

By using -prframe and -sqframe flags, g_sans is going to analyze on each
> frame in the trajectory, I would say the error depends on both (the frames
> read and the frames analyzed).
>
> 2015-03-24 12:07 GMT+01:00 Mark Abraham <mark.j.abraham at gmail.com>:
>
> > Hi,
> >
> > Does the erroneous behaviour depend on the number of frames read, or the
> > number of frames analyzed?
> >
> > Mark
> >
> > On Tue, Mar 24, 2015 at 11:21 AM, CHEN Pan <evan.pan.chen at gmail.com>
> > wrote:
> >
> > > Dear all,
> > >
> > > I am using g_sans_mpi (4.6.4 version) to calculate a SANS curve of a
> > system
> > > containing about 15000 atoms in total (water, urea, NaOH and
> oligomers).
> > > The command used are listed below.
> > >
> > > *echo 0 | g_sans_mpi -f ../md.xtc -s ../md.tpr -pr -sq -pbc -mode
> direct
> > > -xvg none -b 50000 -e 100000*
> > >
> > > The thing is that with calculation for 5 ns of MD trajectory, no error
> > > occurred, whereas for calculation of 50 ns trajectory. An error message
> > > shows " zsh: bus error (core dumped)", or shows "segmentation fault 11"
> > > sometime. With increasing memory or increasing number of threads (-nt)
> > > didn't help.
> > >
> > > I have googled for both "g_sans" and "bus error". It seems nobody has
> > ever
> > > meet such problem with using g_sans. I understand the manual says that
> > > computational cost is going to increase when using -pr and -sq, as well
> > as
> > > the increase of number of system particles. Then I tried to calculate
> > SANS
> > > curves for each frame with the following command. Still, the same error
> > > occurred after reading over 10 ns of frames.
> > >
> > > *echo 0 | g_sans_mpi -f ../md.xtc -s ../md.tpr -prframe -sqframe -pbc
> > -mode
> > > direct -xvg none -b 50000 -e 100000*
> > >
> > > Welcome for any comments and potential explanations.
> > >
> > > Best,
> > > Pan
> > > --
> > > 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.
> >
>
>
>
> --
> Pan Chen
> Postdoctoral Associate
> Tailor-Made Fuels from Biomass
> AICES Graduate School
> RWTH Aachen University
> +49(0)241 8099205
> --
> 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.
>