[QE-users] optimization calculation does not converge within time limit

[Lorenzo Paulatto]

> Taking into consideration what you and Ari said, I changed my input file 
> to this (decreased the number of atoms and decreased the K points). Do 
> you think it will converge?
It's hard to say, just by looking at it. It think you should try to run 
it. I would recommend starting with fewer than 192 CPUs, maybe something 
in the order of 32 with as many parallelization pools an possible (4, if 
your system has no symmetry)

mpirun ... pw.x -npool 4 -in pw.in ...

Sometimes running with too many CPUs is slower. If after half an hour 
with 32 it is running, try with 64 or 128 and see if it goes faster or 
slower.

cheers


-- 
Lorenzo Paulatto - Paris