[Wien] Speedup with ELPA and GPUs

[Straus, Daniel B]

Has anyone tried running WIEN2k linked with ELPA compiled for GPUs? If so, what sort of performance speedup on lapw1 was seen? I get about 1.6 teraflops in lapw1 on a 16 core AMD 9950x using ELPA with 4 OpenMP threads and 4 MPI processes, but some large calculations are still taking weeks.


Daniel Straus
Assistant Professor
Department of Chemistry
Tulane University
5088 Percival Stern Hall
6400 Freret Street
New Orleans, LA 70118
(504) 862-3585
http://straus.tulane.edu/


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20260915/3fcaba1a/attachment.htm>