Depend on openmpi instead of nci-openmpi and add restart reproducibility#49
Depend on openmpi instead of nci-openmpi and add restart reproducibility#49aidanheerdegen merged 5 commits intomainfrom
Conversation
harshula
commented
Feb 21, 2024
- Requires: libaccessom2: force CMake to use MPI compiler wrappers access-spack-packages#66
- Requires: common/gadi/packages.yaml: use system openmpi directly spack-config#26
|
This change was triggered by ACCESS-NRI/access-spack-packages#61 . |
CodeGat
left a comment
There was a problem hiding this comment.
We'll need to update the version before merging it - if you rebase onto main it should help out telling what needs to be done.
* Requires: ACCESS-NRI/access-spack-packages#66 * Requires: ACCESS-NRI/spack-config#26
644da04 to
2ef105b
Compare
|
The model version in the
|
|
!bump minor |
|
✅ Version bumped from |
|
This
It will be deployed using:
If this is not what was expected, commit changes to |
|
Pre-release deployment is reporting
Despite this change Seems that on |
|
Might need to check the version tagged in |
7ce0caa to
2ef105b
Compare
|
The model version in the
|
|
This
It will be deployed using:
If this is not what was expected, commit changes to |
|
!bump major |
|
✅ Version bumped from |
|
This
It will be deployed using:
If this is not what was expected, commit changes to |
|
We shouldn't merge until we have done some performance testing, confirming the impact is minimal |
|
Good point @aidanheerdegen - I've requested changes so it shouldn't be able to merge |
|
Seems the pre-release build is still using |
|
What does this mean, exactly...? |
…i directly instead of nci-openmpi)
|
This
It will be deployed using:
If this is not what was expected, commit changes to |
|
@harshula The openmpi thing was fixed - we hadn't tagged the |
|
Hi @CodeGat , Is there anything else we need to do? |
|
I assume some kind of performance testing of some kind before proper deployment |
|
This pull request has been mentioned on ACCESS Hive Community Forum. There might be relevant details there: |
aidanheerdegen
left a comment
There was a problem hiding this comment.
Looks good. Performance checks suggest no issue with model slowdown.