-
Notifications
You must be signed in to change notification settings - Fork 24
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
include mkl libraries for intel on linux and MacOS #81
Conversation
iulusoy
commented
Mar 29, 2024
- mkl libraries included for most MacOS versions
- mkl libraries included for all linux versions
- new test for mkl installation
- currently is not the default, the mkl test is run conditionally. With the inclusion of math libraries for more compilers+OS, the tests should run as default
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.
to get ci tests to run you can add the math branch to the pull_request
trigger in test.yml
- name: Test MKL libraries | ||
# we could also exclude 2021.5 for all macos versions here | ||
# this needs refactoring at some point - should the install | ||
# fail if no mkl version exists? |
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.
I think it makes sense to fail, given we aim to support all intel versions and runner images barring a few exceptions. In general I think we should fail if we can't satisfy the user's selected configuration.
On the other hand, we don't fail with compiler: intel
on mac, we just fall back to intel-classic
. Maybe there is an argument for changing this.
There is also the question of what to do if the user requests MKL with a non-intel compiler. I would say fail in that case too?
@iulusoy I rebased |
Did so and also added you to the repo. |
thanks @iulusoy I joined. in future you can just check this box to give a repo maintainer access to a PR branch from your fork without making them a full contributor |
@iulusoy can you revert the CSV changes and/or rebase? I updated In general you can ignore the auto-update PRs on your fork (really they should just run on the main repo, I'll change that at some point) and just rebase from upstream instead. |
@iulusoy any idea on the test failures? |
No, not really - I need to try this out in a container to see what is happening. Have it on my to-do list. What about extending the math support to other compilers/windows? Should we assign responsibilities to move this forward? |
I'm still not sure on the approach for other toolchains. Since BLAS/LAPACK were added to This PR could be merged with mac/linux support and windows could come in a separate PR, then Just thinking out loud |
Seems that the apt installation of the mkl libraries changes the setvars script for mkl version 2021.1. Because of this I changed the order of the compiler vs mkl installation. The PR will be updated shortly. |
@wpbonelli do you think this can be merged now, or do we need some more tests? |
@iulusoy do you mind doing a final rebase? looks like the diff still has changes in the compat scripts and CSVs. |
Should be no more diffs to main now, other than the actual changes. |
Looks like this is still behind. Maybe helpful to squash everything down to your first commit e.g. |
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.
looks good to me, merging now.