I am sorry to bother everyone again but there is
a question I can’t understand.
I peformed a MD simulation and PPPM Ewald method
is used to handled the long-range coulomb interactions.
In the PPPM Ewald, the fast Fourier transform (FFT) is
employed and this should cost some time in computing
the long-rang interaction in my opinion.
But, here is the last part of my log.lammps, which
shows the PPP timing information.
PPPM timing info:
Make_rho time = 452.786283392905
Make_rho of long time = 14.4309637768169 Poisson time = 11.3112034606929 Poisson of long time = 0.360504665005105
Electric_field time = 2585.50093744964
Electric_field of long time = 82.4037117327714 Brick2fft time per timestep = 0.000000000000000E+000 Total brick2fft time = 0.000000000000000E+000 Brick2fft of long time = 0.000000000000000E+000
Fillbrick time per timestep = 0.000000000000000E+000
Total fillbrick time = 0.000000000000000E+000
Fillbrick of long time = 0.000000000000000E+000 FFT time per timestep = 0.000000000000000E+000 **Total FFT time = 0.000000000000000E+000** FFT of long time = 0.000000000000000E+000
In light of above info, I can’not understand why the
'Total FFT time ’ is zero.
DO I have something wrong with the Ewald method ?
Thanks a lot.