[gmx-users] mdrun_mpi seg fault if N_atoms/cpu > 4096 ?

Atte Sillanpää atte.sillanpaa at csc.fi
Wed Nov 9 21:48:37 CET 2005

On Wed, 9 Nov 2005, David wrote:

> On Wed, 2005-11-09 at 18:46 +0200, Atte Sillanpää wrote:
>> we have a system with 128 DPPC-molecules and a layer of water. All goes
>> well with version 3.2.1 if the number of atoms per cpu is less than 4096.
>> That is, we get a seg fault before any real md in the beginning:
> This could be the old bug when you have no water on the first processor.
> This has been fixed in 3.3, but you can also use the shuffle option as a
> workaround (will also give you better performance)
Thanks David, you were right and this solved the problem (and likely 
improved performance :)

For the record, without shuffle:

There are 3783 molecules, 8007 charge groups and 17365 atoms
There are 0 optimized solvent molecules on node 0
There are 0 optimized water molecules on node 0

After shuffle there are some waters also on node 0.


More information about the gromacs.org_gmx-users mailing list