[Pw_forum] Fwd: nstep, npool and FFT grid

[Paolo Giannozzi]

On Aug 25, 2007, at 23:06 , brsahu at physics.utexas.edu wrote:

> In a system where the FFT-grids along x-, y- and z-directions are not
> equal,which one of these decides the PW parallelization efficiency?

as mentioned, but not in a clear way, in the user guide, the FFT arrays
are "sliced" in real space along z. The most effective choice of axis
depends on the specific case: if you want to run on as many processors
as possible, it is typically convenient to have the largest FFT grid  
along z.

> Is there a default for number of pools of processors that a running  
> job
> assumes if not specified  in the "job" submitting script

a single pool, i.e. no k-point parallelization

Paolo
---
Paolo Giannozzi, Democritos and University of Udine, Italy