[CPMD-list] test calculation problems.

[Alessandro Curioni]

Yes.
The tests, were run on 32 processors POWER4 in full MPI mode.

The differences (less than 10-4) have to be ascribed mainly to the
different fft mesh used
(154 96 96 versus 160 96 96) imposed by the different FFT libraries used.

BTW if you look in the cpmd-tests distribution, in each subdirectory  you
have a README
that reports the duration and memory needed for each run ( with indication
of the machine used).
README files are written to be read.
Regards.

Alessandro.