New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
GROMACS regression tests fail on Vega with ROCm 1.8 #427
Comments
Here is what we see taccuser@ROCM-REL-VG10:~/Desktop/gromacs/gromacs-2018/build$ make check 100% tests passed, 0 tests failed out of 39 Label Time Summary: Total Test time (real) = 213.14 sec |
This is ROCm 1.8 and newer 1.8.1 we working on |
Yes @pszi1ard , We are not observing any failures as Greg said. |
@gstoner @rkothako Thanks for running the tests. Please help me figure out where is the discrepancy as I can 100% repro failing MdrunTests unit test on my hardware with the latest ROCm from the deb repos:
The same build / binary on a different machine (same OS) with a Fiji passes the test however. Now, as a side-note, unfortunately, due to a bug in our end-to-end regressiontest script, some failing tests don't get reported. Can you also try running them manually, please, i.e.:
|
Thanks @pszi1ard, I am able to reproduce this issue. We logged an internal ticket and working on this. taccuser@ROCM-REL-FIJI:~/Desktop/gromacs/gromacs-2018/build/tests/regressiontests-2018$ perl gmxtest.pl -xml complex -nt 1
Aldert van Buuren Rudi van Drunen Anton Feenstra Gerrit Groenhof Copyright (c) 1991-2000, University of Groningen, The Netherlands. GROMACS is free software; you can redistribute it and/or modify it GROMACS: gmx mdrun, version 2018 Thanx for Using GROMACS - Have a Nice Day Abnormal return value for ' gmx mdrun -ntmpi 1 -notunepme >mdrun.out 2>&1' was -1 Abnormal return value for ' gmx mdrun -ntmpi 1 -notunepme >mdrun.out 2>&1' was -1 Abnormal return value for ' gmx mdrun -ntmpi 1 -notunepme >mdrun.out 2>&1' was -1 Abnormal return value for ' gmx mdrun -ntmpi 1 -notunepme >mdrun.out 2>&1' was -1 Abnormal return value for ' gmx mdrun -ntmpi 1 -notunepme >mdrun.out 2>&1' was -1 Abnormal return value for ' gmx mdrun -ntmpi 1 -notunepme -cpi ./continue -noappend >mdrun.out 2>&1' was -1 |
Thanks for checking, please let me know if I can assist in fixing this issue. |
Multiple regressiontests fail on ROCm 1.8 while these do pass with AMDGPU-PRO.
To reproduce follow the "Quick and dirty" installation instructions and the
make check
stage should reveal the issues: http://manual.gromacs.org/documentation/2018/install-guide/index.html#quick-and-dirty-installation.The text was updated successfully, but these errors were encountered: