[gmx-users] Fwd: Parallellization problem

[Matthew Zwier]

I've had bad luck with parallel minimizations, particularly for the
4.0 series of GROMACS.  Either domain decomposition fails or numeric
problems appear (SETTLE failures and the like), but disappear when run
serially.  Minimization tends to be low cost compared to equilibration
anyway, so my solution has simply been never to run a minimization in
parallel, even for my large systems.  They're still done in minutes or
hours, compared to the hours or days required for equilibration.

On Mon, Jun 20, 2011 at 11:41 AM, Nuria Alegret <nuria.alegret at urv.cat> wrote:
>
>  Good afternoon,
>
> I'm using GROMACS version 4.0.5. My simulation system is a double stranded
> DNA (51 nucleotides) in a water (TIPI3P type) box, defined as 0.9 A from the
> DNA, with 100 Na ions to neutralize the system. The sequencial commands used
> were:
>
> 1) pdb2gmx -f dsDNA.pdb -p dsDNA.top -o dsDNA.gro -ffamber99
>
> 2) editconf -f dsDNA.gro -o box.gro -d 0.9
>
> 3) genbox -cp box.gro -cs ffamber_tip3p.gro -o water.gro -p dsDNA.top
>
> 4) grompp -f em.mdp -c water.gro -p dsDNA.top -o Premin.tpr
>
> 5) genion -s Premin.tpr -o water-ions.gro -np 100
>
> 6) grompp -f em.mdp -c water-ions.gro -p dsDNA.top -o minIons.tpr
>
> 7) MINIMIZATION:  mdrun -v -s minIons.tpr -o minIons_traj.trr -x
> minIons_traj.xtc -c minIons_final.gro -e minIons_ener.edr
>
>
> The problem appears when I try to minimize the system. When I try to
> parallellize the calculation, the calculation returns with an error message
> as follows:
>
> Fatal error:
> There is no domain decomposition for 8 nodes that is compatible with the
> given box and a minimum cell size of 38.1122 nm
> Change the number of nodes or mdrun option -rdd
> Look in the log file for details on the domain decomposition
>
> According to your web page, this error appears when a very small system is
> too small to run parallellized. But my system is extremelly big!! In
> previous calcs I could parallellize smaller systems, even with just 8
> nucleotides.
> I attach you the error file and the log file, as well as the mdp.
> I would be gratefull with any help. Thanks in advance,
>
>
>
> Núria Alegret
> University Rovira i Virgili - Tarragona (Spain)
>
>
>
> --
> 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
>