-
Notifications
You must be signed in to change notification settings - Fork 23
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
Switch to CTest for CI and add NVHPC 23.5 to CI testing #56
Conversation
851bd85
to
ef46b1a
Compare
ef46b1a
to
852c087
Compare
fffdd81
to
bf5c149
Compare
bf5c149
to
89cd907
Compare
c8e0273
to
0749717
Compare
0749717
to
df29dc8
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Changes look neat, and seem to reduce the test matrix quite significantly. As far as I can see there is no loss of functionality (only some error norm logging) that gets lost in using ctest, so all good from me, and GTG.
run: sudo apt-get install libopenmpi-dev | ||
|
||
# Free up disk space for nvhpc | ||
- name: Free Disk Space (Ubuntu) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
😮
Original intention was just to add a newer NVHPC version to Github CI but it went sideways pretty badly. Reason being that NVHPC seems to apply unsafe vectorization code paths that fail on the tail block when running with larger NPROMA values. Catering for too many different launch configurations in the runner script becomes unnecessarily complicated, since known good configurations are already encoded as ctests in the CMake configuration. Therefore, I've switched the CI runs to use CTest now (but kept the target validation in place).