[gmx-users] g_tune_pme

[Carsten Kutzner]

Hi Carla,

On Jul 28, 2011, at 9:38 AM, Carla Jamous wrote:

> Hi everyone, please I was running simulations with gromacs version 4.0.3 ,but I got the following error:
> Average load imbalance: 12.1 %
>  Part of the total run time spent waiting due to load imbalance: 6.9 %
>  Steps where the load balancing was limited by -rdd, -rcon and/or -dds: X 0 % Y 9 %
>  Average PME mesh/force load: 0.807
>  Part of the total run time spent waiting due to PP/PME imbalance: 5.3 %
> 
This is not an error but just a hint how you could optimize your performance.

> NOTE: 6.9 % performance was lost due to load imbalance
>       in the domain decomposition.
> 
> NOTE: 5.3 % performance was lost because the PME nodes
>       had less work to do than the PP nodes.
>       You might want to decrease the number of PME nodes
>       or decrease the cut-off and the grid spacing.
> 
> After searching the archive mailing list and reading the manual , I decided to use g_tune_pme so I switched to gromacs 4.5.4. Here's my script:
Note that there is also a g_tune_pme version for 4.0.7: 
http://www.mpibpc.mpg.de/home/grubmueller/projects/MethodAdvancements/Gromacs/index.html

As another possibility, you can use the tpr file you created with 4.0.x as input for
Gromcas 4.5.x, also for g_tune_pme, this is probably the easiest solution.

> 
> #PBS -S /bin/bash
> #PBS -N job_md6ns
> #PBS -e job_md6ns.err
> #PBS -o job_md6ns.log
> #PBS -m ae -M carlajamous at gmail.com
> #PBS -l select=2:ncpus=8:mpiprocs=8
> #PBS -l walltime=024:00:00
> cd $PBS_O_WORKDIR
> export GMXLIB=$GMXLIB:/scratch/carla/top:.
> module load gromacs
> chem="/opt/software/SGI/gromacs/4.5.4/bin/"
> mdrunmpi="mpiexec /opt/software/SGI/gromacs/4.5.4/bin/"
> ${chem}grompp -v -f md6ns.mdp -c 1rlu_apo_mdeq.gro -o 1rlu_apo_md6ns.tpr -p 1rlu_apo.top
> ${mdrunmpi}g_tune_pme -v -s 1rlu_apo_md6ns.tpr -o 1rlu_apo_md6ns.trr -cpo state_6ns.cpt -c 1rlu_apo_md6ns.gro -x 1rlu_apo_md6ns.xtc -e md6ns.edr -g md6ns.log -np 4 -ntpr 1 -launch
> 
> But now, I have the following error message: 
> 
> Fatal error:
> Library file residuetypes.dat not found in current dir nor in your GMXLIB path.
Why don't you build your tpr file on your workstation and then switch over
to the cluster? I guess this will make life easier for you.

Also note that you must not call g_tune_pme in parallel (which you do by
${mdrunmpi}g_tune_pme. g_tune_pme will spawn its own MPI processes with the
help of the MPIRUN and MDRUN environment variables. See g_tune_pme -h,
probably you need do set
export MDRUN=/opt/software/SGI/gromacs/4.5.4/bin/mdrun
export MPIRUN=`which mpiexec`

Hope that helps,
  Carsten

> 
> Except that I'm using amber94 force-field and that my topology files are in a special directory called top where I modified certain things. With gromacs 4.0.3, it always worked so I don't know what is happening here.
> 
> Please does anyone have an idea of what it might be?
> 
> Do I have to run pdb2gmx, editconf, etc... with the gromacs 4.5.4 for it to work?
> 
> Thank you,
> 
> Carla
> -- 
> gmx-users mailing list    gmx-users at 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-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/Support/Mailing_Lists


--
Dr. Carsten Kutzner
Max Planck Institute for Biophysical Chemistry
Theoretical and Computational Biophysics
Am Fassberg 11, 37077 Goettingen, Germany
Tel. +49-551-2012313, Fax: +49-551-2012302
http://www.mpibpc.mpg.de/home/grubmueller/ihp/ckutzne