[QE-users] Large input files and pw.x

[Lorenzo Paulatto]

On 1/18/21 9:56 PM, Aaron Friesz wrote:
> That cleared that particular issue right up.  Is there any general 
> info on running large cells that might shed light on the other 
> problems and bottlenecks you hinted at?

The computational cost of DFT (plane-wave based and local-density) 
scales with the cube of the number of electrons, this should allow you 
to estimate the computational cost of your calculation. Depending on 
what your system actually actuallt is (i.e. a large molecule, or a large 
crystal supercell), it may be more effective to use a  code that can use 
localized basis set (like SIESTA) or linear-scaling algorithms (like BigDFT)

cheers