-
Notifications
You must be signed in to change notification settings - Fork 768
{lang,toolchain}[lfbf/2025b,system/system] lfbf v2025b, SciPy-bundle v2025.07 #24482
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
base: develop
Are you sure you want to change the base?
Conversation
… and patches: numpy-2.3.2_fix-llvm-tests.patch
Updated software
|
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
|
With FlexiBLAS default ( Test report by @Thyre |
|
With FlexiBLAS default ( Test report by @Thyre |
This comment was marked as outdated.
This comment was marked as outdated.
|
Failed due to BLIS not knowing about Test report by @Thyre |
|
With Test report by @Thyre |
|
With FlexiBLAS default ( Test report by @Thyre |
Signed-off-by: Jan André Reuter <[email protected]>
Signed-off-by: Jan André Reuter <[email protected]>
This comment was marked as outdated.
This comment was marked as outdated.
|
With FlexiBLAS default ( Test report by @Thyre |
Signed-off-by: Jan André Reuter <[email protected]>
Signed-off-by: Jan André Reuter <[email protected]>
471b89a to
e05de7f
Compare
This comment was marked as outdated.
This comment was marked as outdated.
Signed-off-by: Jan André Reuter <[email protected]>
|
With FlexiBLAS default ( Test report by @Thyre |
|
will also need |
|
Hmm, when testing this on Does it even make sense to build that on an Arm system? edit: that was with a naive " |
@boegel I'm seeing the same failure on our GH200 systems. As we're filtering AOCL-BLAS from aarch64, even in our GCC FlexiBLAS variant, I'm guessing that we hit a similar failure there. I wouldn't spend to much time on it, given that this is the AMD optimized BLIS variant 😅 I've filtered this EC when testing on GH200. |
|
I see two failing tests on @Thyre Does this ring any bells for you? I don't think this should block this PR, we can open a follow-up PR on it |
That's the same test failure I saw here #24482 (comment). |
Make crackfortran.py actually aware of the selected_real_kind of LLVM flang (and Fortran 2018). This doesn't fix aarch64 and Zen 4 test failures later on though. Signed-off-by: Jan André Reuter <[email protected]>
@boegel The test failure may be related to another OpenBLAS bug. Trying the same tests with I've tested this with the latest |
This comment has been minimized.
This comment has been minimized.
|
Test report by @boegel |
We need easybuilders/easybuild-easyblocks#3855 here for Boost 😄 |
|
Test report by @boegel |
|
With Test report by @Thyre |
|
With Test report by @Thyre |
|
Test report by @boegel |
|
Tested with FW PR at easybuilders/easybuild-framework@232ef0b Test report by @Crivella |
|
NOTE Only tested ECs from #24516 Test report by @Crivella |
|
Test report by @boegel |
|
I'm seeing trouble here with BLIS on Test report coming up. I won't let this block this PR though, we can follow up in an issue & separate PR, since this smells like a bug in BLIS to me... |
|
Test report by @boegel |
|
Test report by @Thyre |
Interesting, this doesn't show up with Neoverse V2. Probably worth opening an issue for to figure out where things go wrong. |
(created using
eb --new-pr)This is part of a series of PRs adding an LLVM toolchain for
2025b.Right now, the modules use
llvm-compilers, whereas the framework PR is usingLLVMtc. I'll adapt based on the result of the discussion in Slack.Requires:
Also contains
llvm-compilersto that this can be tested independently from #24481Test reports: