Why is the parallel running not working properly in my script? K space? or anything?

Dear All

As I doubled the number of processors, the memory usage per processor is almost halved. This was true when I run the provided example file in.melt using our cluster.

However, the parallel running failed with script in.BiTiO. When I doubled my number of processors (from 12 processors to 24 processors), the memory usage per processor is almost constant (40 mb). As a result, I cannot run much larger systems.

I suspect it maybe due to the kspace or some missing package that prevent running parallel properly. Wish someone could give some hints and help.

Thanks in advance

in.BiTiO (1.33 KB)

tempfile1 (44 Bytes)

lammps.pbs (387 Bytes)

Kspace Ewald at a precision of 1.0e-6 probably
uses a very large # of Kspace vectors. Each
proc has copy of all of them. Try PPPM,
which will be faster and less memory.