I'm planning to run a large MD simulation of disordered structure using
PPPM to handle the long rang electrostatic. My system has the following
dimensions: 65 x 65 x 197 Angstroms.
My simulation can run easily smoothly on 2 processors with
kspace_style pppm 1.0e-05
When I increase the number of processors, I get into trouble and can no
longer have the simulation to run. I tried to introduce
kspace_modify 120 120 240 order 7
without any success. The whole thing works fine with
kspace_style ewald 1.0e-05
whatever the number of processors I specify. I would like to take
advantage of the speed of pppm over ewald and I desperately need help.