[Q-e-developers] Improved Communication in exx.f90

[Filippo SPIGA]

Hi Taylor,

I committed your patch in the legacy repository. More expert colleagues than me will verify that everything works well. Thank you very much for your contribution !!!

Cheers

--
Mr. Filippo SPIGA, M.Sc.
Quantum ESPRESSO Foundation
http://www.quantum-espresso.org ~ skype: filippo.spiga

*****
Disclaimer: "Please note this message and any attachments are CONFIDENTIAL and may be privileged or otherwise protected from disclosure. The contents are not to be disclosed to anyone other than the addressee. Unauthorized recipients are requested to preserve this confidentiality and to advise the sender immediately of any error in transmission."

> On Feb 1, 2016, at 7:01 AM, Taylor Barnes <tbarnes at lbl.gov> wrote:
> 
> Dear All,
> 
>    I am a postdoc who is currently working on QE as part of NERSC's NESAP efforts, with particular interest in improving the efficiency of the exact exchange parts of the code.  In the coming months I intend to contribute several changes to the parallelization strategy of exx.f90, but in the meantime I would like to present one small modification to vexx_k.
>    The attached patch reduces the total amount of data communication in subroutine vexx_k.  This is accomplished by moving several "mp_sum" calls to a later point in the code, so that quantities that were communicated in real-space are instead communicated in g-space.  For some calculations, this can result in a non-negligible reduction in walltime.  The patch can be applied by running "patch < exx.patch" within PW/src.
> 
> Sincerely,
> Dr. Taylor Barnes
> Postdoctoral Scholar,
> Lawrence Berkeley National Laboratory
> <exx.patch>_______________________________________________
> Q-e-developers mailing list
> Q-e-developers at qe-forge.org
> http://qe-forge.org/mailman/listinfo/q-e-developers