Ilp64 openblas #20
Closed
Ilp64 openblas #20
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Co-authored-by: Christian Lorentzen <[email protected]>
TST: optimize: add test for VectorFunction with sparse_jacobian=False
This makes tests pass under pytest-run-parallel; the logic was already broken for use of the default build with Python-level threads (e.g., through the `threading` module) though. This is per-module global state that wasn't robust against altering it from separate threads.
Also tweak the refcounting test's skip condition to be more accurate. It would have failed under Pyston and other interpreters than CPython and PyPy otherwise.
It's always defined for Python >=3.7, threadless builds no longer exist.
`qhull.pyx` uses cython_lapack, so linking is done at the build time. NB: and if anything, Qhull might not be 64-bit int ready, so it'll probably need an LP64 LAPACK.
Violations observed on linux with MKL in the ILP64 mode.
It otherwise segfaults. Needs using BLAS_FUNC maybe?
…hon_lapack" This reverts commit 8b60d82.
…ellip_harm" This reverts commit 043a40e.
32-bit LAPACK builds generate the `blas_lapack_wrapper.so` library which contains trampolines from F_FUNC to BLAS_FUNC mangling (i.e. create dgemm_ as a synonym for scipy_dgemm_ etc). Then f2py uses F_FUNC mangled symbols, and which get picked up from the wrappers. With ILP64 builds this approach may lead to symbol clashes: at least the MKL _ilp64.so variants contain two kinds of symbols: both `dgemm_64_` and `dgemm_`. We want to only use the former (is dgemm_ 32- or 64-bit?). Therefore, we instead force the BLAS_FUNC mangling at the f2py level: inject a 'usercode' section into f2py-generated C files and replace the F_FUNC define with BLAS_FUNC. This relies on the fact that f2py includes the usercode section _after_ its F_FUNC define.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Reference issue
What does this implement/fix?
Additional information