[gmx-users] Avoiding domain decomposition error

[Mark Abraham]

Sure, using mpirun -np x mdrun_mpi will check for this. You don't want to
actually compute on the head node (your sysadmins know where I live!), but
all you need to do to trigger the check is to spawn an MPI job with that
many ranks, and then kill it immediately if there was no error. (Not sure
if adding -nsteps 0 will get the job done, but -nsteps 1 certainly
will.) You don't need those ranks to be spawned across a real compute
system for this test - over-subscribing any machine is fine.

Mark


On Tue, Nov 19, 2013 at 5:53 PM, Ricardo Oliveira dos Santos Soares <
rsoares at fcfrp.usp.br> wrote:

> Hello fellow users,
> is there a direct way of checking the domain decomposition viability
> before running a job, thus avoiding wasting large queue times on a shared
> cluster only to receive the attached error?
> Is this something that can be done with aid of g_tune_pme?
>
> Here's the link to this well know issue:
>
> http://www.gromacs.org/Documentation/Errors#There_is_no_domain_decomposition_for_n_nodes_that_is_compatible_with_the_given_box_and_a_minimum_cell_size_of_x_nm
>
> Thanks,
>
> Ricardo.
>
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>