Failing tests in Gromacs 2020.6 and 2021.2 with Ryzen9 5950X

have just mount a new workstation with a AMD Ryzen 9 5950X cpu and Geforce GTX1650 video card. The installed OS was Opensuse Leap 15.2,

The main purpose of this workstation is to rum simulations with Gromacs. After its compilation without any warning, I’m having the following tests failing when checking it:

32 - RandomUnitTests (Failed)
40 - GmxAnaTest (Failed)
41 - GmxPreprocessTests (SEGFAULT)
48 - TrajectoryAnalysisUnitTests (Failed)
50 - ToolUnitTests (SEGFAULT)
53 - MdrunOutputTests (Failed)
54 - MdrunModulesTests (Failed)
55 - MdrunIOTests (Failed)
56 - MdrunTests (Failed)
57 - MdrunPmeTests (Failed)
60 - MdrunMpiTests (Failed)
61 - MdrunMpiPmeTests (Failed)
64 - MdrunFEPTests (Failed)
66 - GmxapiExternalInterfaceTests (Failed)
67 - GmxapiInternalInterfaceTests (Failed)

Essentially two kinds of failure appear in the tests:

C++ exception with description “random_device: rdrand failed”

and

Attempted to call init_inputrec_strings before calling done_inputrec_strings.

The first failure above, the rdrand one, appears to be related with some AMD CPU’s, as described here:

https://twitter.com/FiloSottile/status/1125840275346198529 1

I made everything possible to try to fix these failures by my self, including the compilation and instalation of GCC 11.1.0 and Intel oneAPI toolkit and use them to compile gromacs 2021.2. Every time I finished with the same errors. :-(

changed the description

changed title from Failing testes in Gromacs 2020.6 and 2021.2 with Ryzen9 5950X to Failing tests in Gromacs 2020.6 and 2021.2 with Ryzen9 5950X

changed milestone to %2022.beta2

More RDRAND issues :-/ AMD really dropped the ball on this one.

I'm adding this to the beta2 milestone; it seems a simple solution is to just use getrandom() on Linux at least, and hopefully the suspend functionality on windows is different.

That at least one std library now detects this and fails to create a useless random device is a Good Thing. GROMACS dropped the ball in not being prepared to catch the documented exception.

mentioned in commit 9c2b752d

mentioned in commit 9c06c020

mentioned in merge request !2147 (merged)

Hi @parpires!

Could you please check if the patch in !2147 (merged) (9c06c020) for GROMACS 2021 fixes the issue for you?

Dear Andrey, Would you please explain how I can compile Gromacs 2021 with the above mentioned patch? Thank you very much.

Paulo A.

...

Em ter., 2 de nov. de 2021 às 12:14, Andrey Alekseenko (@al42and) < gitlab@mg.gitlab.com> escreveu:

Andrey Alekseenko https://gitlab.com/al42and commented #4049 (comment 721235577):

Hi @parpires https://gitlab.com/parpires!

Could you please check if the patch in !2147 (merged) !2147 (merged) (9c06c020 9c06c020) for GROMACS 2021 fixes the issue for you?

— Reply to this email directly or view it on GitLab #4049 (comment 721235577). You're receiving this email because you have been mentioned on gitlab.com. If you'd like to receive fewer emails, you can unsubscribe https://gitlab.com/-/sent_notifications/REDACTED/unsubscribe from this thread or adjust your notification settings.

-- Paulo Augosto Rodrigues Pires Doutor em Química e Especialista em Lab. Lab. Prof. Omar A. El Seoud Instituto de Química da USP Tel. +55-11-30913864

You can get a tarball of the updated code from the page of the branch associated with !2147 (merged). For your convenience, that's https://gitlab.com/gromacs/gromacs/-/tree/aa-4049-failing-tests-in-gromacs-2020-6-and-2021-2-with-ryzen9-5950x. It's not quite the same as a real GROMACS source tarball, but you can compile it the same way you usually do.

Dear Mark, I compiled Gromacs on the computer with Ryzen 9 5950x and ran make check. It ran all tests without any errors! :-)) That computer is running OpenSuse Leap 15.2 and the used compiler was gcc version 11.1.0. I´m sending, attached, the results of the tests.

result_tests.txt

...

Em ter., 2 de nov. de 2021 às 19:10, Mark Abraham (@mark.j.abraham) < gitlab@mg.gitlab.com> escreveu:

Mark Abraham https://gitlab.com/mark.j.abraham commented on a discussion #4049 (comment 721604239):

You can get a tarball of the updated code from the page of the branch associated with !2147 (merged) !2147 (merged). For your convenience, that's https://gitlab.com/gromacs/gromacs/-/tree/aa-4049-failing-tests-in-gromacs-2020-6-and-2021-2-with-ryzen9-5950x. It's not quite the same as a real GROMACS source tarball, but you can compile it the same way you usually do.

— Reply to this email directly or view it on GitLab #4049 (comment 721604239). You're receiving this email because of your account on gitlab.com. If you'd like to receive fewer emails, you can unsubscribe https://gitlab.com/-/sent_notifications/REDACTED/unsubscribe from this thread or adjust your notification settings.

-- Paulo Augosto Rodrigues Pires Doutor em Química e Especialista em Lab. Lab. Prof. Omar A. El Seoud Instituto de Química da USP Tel. +55-11-30913864

Great, thanks! The fix will be in 2021.4. Thanks for the bug report and sorry about the long time to deal with it!

assigned to @al42and

mentioned in commit 1ca44a96

changed milestone to %2021.4

closed