[gmx-users] g_tcaf issue

[Justin Lemkul]

On Fri, Dec 13, 2013 at 2:01 PM, Jones de Andrade <johannesrs at gmail.com>wrote:

> Hi all.
>
> We are having some problem running g_tcaf here. It keeps yielding
> "segmentation faults".
>
> We first thought it was because of the single-precision calculations
> of the first run, or the exceedingly large trajectory file. Both were
> changed for a second trial, which still yielded the same error.
>
> Also, we are aware of the similar error posted in this thread at this
> mail-list:
>
> http://www.mail-archive.com/gmx-users@gromacs.org/msg54295.html
>
> Unfortunately, it seems to be a different issue.
>
> We are also writing a full trajectory file in the .trr file, as can be
> seen below on the excerpt from the .mdp file below:
>
> dt                  =  0.002
> nsteps              =  100000
> nstcomm             =  1
> nstxout             =  0
> nstvout             =  10
> nstfout             =  10
> nstlog              =  10
> nstenergy           =  100
> nstlist             =  5
>
> The precise line our cluster script is trying to execute is:
>
> echo 0 | g_tcaf -f BMIm.AlCl4.md14.trr -s BMIm.AlCl4.md14.tpr -oa
> BMIm.AlCl4.md14.all.xvg -o BMIm.AlCl4.md14.tcaf.xvg -of
> BMIm.AlCl4.md14.fit.xvg -ov BMIm.AlCl4.md14.visc.xvg >& md14.tcaf.out
>
> The "echo 0" assures that the needed parameter is passed to g_tcaf. By
> the way, in our last test the pbs parameters for resources request
> were:
>
> #PBS -l ncpus=8
> #PBS -l mem=16GB
>
> We know that it's pointless to ask for extra cpus, it's in there just
> for a matter involving the machine architecture. But the 16 GB of
> memory, despite available, are just our guess on how much it will
> need. We also tested it out of the queue to be certain, but it still
> fails at the same point.
>
> Does anybody has any suggestion please? I would be really helpful.
>
>
Seg faults from analysis tools are usually bugs.  Feel free to file a bug
report on redmine.gromacs.org with the smallest test case possible that
reproduces the problem.  Running the command through a debugger like gdb
would help to find the solution even faster.

-Justin

-- 

==========================================

Justin A. Lemkul, Ph.D.
Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441


==========================================