I am working on HP workstation with :
- 128 GB of RAM
- INTEL(R) XEON E5-2620 V3 with 12 cores
What are the best settings of NCORE, KPAR, etc. to speed up calculations on this computer?
$\begingroup$
$\endgroup$
5
-
3$\begingroup$ I am not a VASP user but I would guess that the optimal parallelization scheme also depends on each calculation you want to perform. $\endgroup$– Gregor MichalicekMar 27, 2021 at 17:06
-
3$\begingroup$ kindly check this link which contain benchmarking link for vasp mattermodeling.stackexchange.com/questions/4525/… $\endgroup$– Pranav kumarMar 27, 2021 at 17:28
-
1$\begingroup$ Use Intel compilers, they are the best on Intel processors. $\endgroup$– S R MaitiMar 27, 2021 at 20:41
-
1$\begingroup$ This question is currently being discussed in the VASP chat room. $\endgroup$– Nike DattaniMar 28, 2021 at 0:55
-
1$\begingroup$ I think this would constitute a duplicate of the question linked by pranav kumar. $\endgroup$– Tyberius ♦Apr 12, 2021 at 16:07
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
2
How to choose these parameters is dependent on your system, which may need to test to find the most optimized computational strategy.
Here is a good tutorial to talk about these setting: https://www.nsc.liu.se/support/Events/VASP_workshop_2020/seminar3.pdf
- NCORE:
NPAR:
KPAR:
-
1$\begingroup$ I expect this question will be closed as a duplicate. You can leave your answer here, but it may be more visible if you move it to the question linked by pranav kumar. $\endgroup$– Tyberius ♦Apr 12, 2021 at 16:08
-
1$\begingroup$ @Tyberius You mean I should put this answer to that post that Pranav Kumar linked? :) $\endgroup$– JackApr 13, 2021 at 1:35