[gmx-users] mdrun -rdd and -dds flag

[Broadbent, Richard]

On 11/04/2013 11:17, "manara r. (rm16g09)" <rm16g09 at soton.ac.uk> wrote:

>Dear gmx-users,
>
>I am having a problem with a periodic molecule and the domain
>decomposition, I wish to use a high number of processors (circa 180, but
>can obviously be reduced) and therefore need to use the -rdd or -dds
>flags (I believe), how do these value effect the simulation?
This is well documented and can be easily viewed by running

mdrun -h

In general these will be set automatically and do not need to be altered.
 
>
>The box size is 11.5 x 11.5 x 45.
>
>The polymer contains charged groups every repeating unit.

Depending on how these are setup this might be where your problem lies.
Charge groups are generally be small neutral groups of atoms say CH3 and
CH2 in a polyethylene molecule or CH in benzene. If your monomer is large
and treated as 1 charge group that could cause some problems. The exact
setup depends on how they were parameterised in the forcefield. If you are
using gromacs 4.6 or 4.6.1 with the verlet-cutoff charge groups have no
effect anyway.


>
>The error I'm getting is (note this is a test run on my home machine, not
>the cluster):
>Fatal error:
>There is no domain decomposition for 4 nodes that is compatible with the
>given box and a minimum cell size of 61.7923 nm
>Change the number of nodes or mdrun option -rdd or -dds

Since that is larger than your box size it suggests something is
dramatically wrong with your setup. The two things I know of that could
potentially cause this are strange constraints in the topology combined
with LINCS, or inappropriate choice of charge groups in the topology. My
guess is the latter.

>
>Regards
>
>Rich Manara
>PhD Student
>Chemistry
>Southampton University
>UK
>-- 
>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