Questions regarding the compilation of VASP on various platforms: hardware, compilers and libraries, etc.
Moderators: Global Moderator, Moderator
-
hszhao.cn@gmail.com
- Full Member
- Posts: 191
- Joined: Tue Oct 13, 2020 11:32 pm
#1
Post
by hszhao.cn@gmail.com » Tue Feb 22, 2022 3:02 pm
Dr. Fortran Steve Lionel told me the
following:
If the VASP validation test is looking for an exact FP value, that’s its error. If it accepts a small variation, perhaps the range needs to be widened a bit to accommodate more advanced vectorizations.
So I want to know the tolerance used by the testsuite, and wonder if the tolerance range needs to be widened a bit to accommodate more advanced vectorizations, say, AVX512 and similar scenarios.
Regards,
HZ
-
ferenc_karsai
- Global Moderator
- Posts: 460
- Joined: Mon Nov 04, 2019 12:44 pm
#2
Post
by ferenc_karsai » Tue Mar 01, 2022 7:14 am
One has to be very cautious when widening the tolerance for the testsuite. Test suites should be as "universal" as possible so it's often better to tweak the parameters and make the affected examples more "stable".
All the tolerances are set in testsuite/runtest.