[gmx-users] parallel PME documentation request

[Mark Abraham]

Hi,

While writing some scripts that iterated over PME parameters looking to 
satisfy some optimality criteria, I ran into a problem that grompp needs 
fourier_nx to be divisible by the number of nodes in order to construct 
the .tpr file. This isn't noted anywhere in the documentation, and it 
being there would have saved me an hour or two of diagnosis of the 
problem. If anywhere, I think the section on Run Parameters for Ewald 
would be best.

Regards,

Mark