|
Thanks for your reply.
I had set DFLAGS and autotools did not update it, therefore, "-D__ELPA" was missing during the compilation of LAXlib.
Now it seems to be fine and I see ELPA in the output, instead of
scalapack, in line: "ELPA distributed-memory algorithm ......"
However, the performance of QE with ELPA is not better than QE with scalapack. The system has more than 2,000 KS orbitals running on more than 500 MPI processes. Also, I played around with "-nd" and "-nt" (the latter for FFT), they have some significant impact on run time but the same for both ELPA and scalapack.
Do QE developers recommend users to use ELPA? Is there a benchmark of QE comparing its performance with and without ELPA? ***************
On 4/15/23 10:08, Paolo Giannozzi
wrote:
On 13/04/2023 19:28, Alireza Ghasemi wrote: -- Dr. S. Alireza Ghasemi Training & Support Erlangen National High Performance Computing Center Friedrich-Alexander-Universität Erlangen-Nürnberg Martensstrasse 1, 91058 Erlangen, Germany https://hpc.fau.de/about-us/people |
_______________________________________________ The Quantum ESPRESSO community stands by the Ukrainian people and expresses its concerns about the devastating effects that the Russian military offensive has on their country and on the free and peaceful scientific, cultural, and economic cooperation amongst peoples _______________________________________________ Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list [email protected] https://lists.quantum-espresso.org/mailman/listinfo/users
