On 7/09/2011 1:01 PM, Nathalia Garces wrote:
Hello,
The last 6 months I've been working with gromacs (4.5.3 and 4.5.4) in
Ubuntu (10.4) and it has worked fine. The problem started when I
changed the linux distribution to Scientific Linux (5.3)... I followed
the same steps to download/configure gromacs but something's wrong or
missing because the simulations that I had been running on UBUNTU
crashed in SL.
Maybe you just got lucky. The success of poorly-conditioned numerical
integration procedures can vary with what the user ate for lunch. Well,
not quite, but you get the idea. :)
You can make a better assessment by transferring a fully equilibrated
.tpr from your working machine to your non-working machine and seeing
whether that runs.
The command I use is very simple:
/ mdrun -v -s topol.tpr/
In Ubuntu the simulation ends fine but in SL the application starts to
write on prompt at early steps (around 100)
/Step XX, time 0.XX (ps) LINCS WARNING/
/relative constraint deviation after LINCS:/
/rms XX, max XX (between atoms XX and XX)/
/bonds that rotated more than 30 degrees:/
/.../
//Wrote pdb files with previous and current coordinates//
/
/
Until the simulation ends (around 250) with a message of
/
Segmentation fault
/
I tried different ways to configure gromacs but none of them seems to
resolve the problem
1) ./configure --prefix=<my path>
make
make install
2) ./configure --with-fft=fftw3 --prefix=<my path> --without-x
make
make install
3)./configure --program-suffix= --disable-float --prefix=<my
path> --exec-prefix=<my path> --enable-all-static --with-fft=fftw3
--without-x
make
make install
You will need an FFTW install with float/double precision matching that
of the GROMACS configure command.
Moreover, I downloaded the benchmark posted in the gromacs web page to
test the simulations with other ".tpr" files. I run the first
simulation (implicit solvent) in UBUNTU and worked OK but the one on
SL crashed (around step 460) with the first two options of
configuration. The last configuration enabled the simulation to run
OK( I decided to stop it at 4000 steps)...
This bewilders me because the last configuration permitted the
benchmark to run OK but not my file. I don't think my file has an
unstable system because it works all right in UBUNTU (and also Debian!)
One possibility is some kind of dynamic linking failure from all your
different installations. If you always install to a --prefix that you
can delete later, then you can be sure that this kind of problem cannot
occur when you are testing different installations.
In addition, I upgraded my linux distribution to the latest SL 5.x
release but it didn't make any change!
I hope you can help me with some guide on how to proceed a GOOD
instalation in SL or to let me know what is my problem and what I am
missing.
Be sure you are following all the advice on the installation page of the
GROMACS webpages.
Mark
Thank you for four answer,
Nathalia Garces
--
gmx-users mailing list gmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
Please don't post (un)subscribe requests to the list. Use the
www interface or send it to gmx-users-requ...@gromacs.org.
Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
