[gmx-users] grompp allocating 0 atoms to a node

[Justin A. Lemkul]

Hi all,

I'm running a very simple system consisting of a globular protein with a ligand
in a box of SPC water with counterions (ffG43a1 for all components).  I ran my
normal routine of minimization, equilibration, and production, but I've been
getting poor scaling and decided to try again with a different number of
processors.  The system contains roughly 39,000 atoms, and I initially used 24
processors.  Now I would like to try 36.  When I run grompp, it allocates 0
atoms to the first node:

Division over nodes in atoms:
       0    3135    1059    1059    1059    1059    1059    1059    1059    1059
   1059    1059    1059    1059    1059    1059    1059    1059    1059    1059 
  1059    1059    1059    1056    1059    1056    1059    1056    1059    1056  
 1059    1056    1059    1056    1059    1056

I tried the -shuffle option, just to see if it would change anything, although I
didn't expect that it would do much since my system is so simple.

The Gromacs version is 3.3, and I have not tried newer versions for
consistency's sake with some older simulations we have done.

Has anyone ever experienced anything like this?  Is there anything wrong?  It
seems odd to allow zero atoms to be on one node.

Thanks for any insights,
Justin


========================================

Justin A. Lemkul
Graduate Research Assistant
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul at vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/

========================================