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
Make ol_compatible_view accessible on all targets #8537
Conversation
This is to fix Issue numba#8529, where test_reinterpret_array_type fails on CUDA with NumPy 1.23 because this overload is needed. Making it accessible to the CUDA target should resolve the issue.
gpuci run tests |
/azp run |
Azure Pipelines successfully started running 1 pipeline(s). |
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.
Thanks for the patch. This has been manually tested locally as part of debugging: #8529 (comment). I think it also makes sense as the most minimally invasive method to resolve the problem, i.e. the CUDA target using some lowering from the CPU target that calls out to an @overload
... that @overload
needs to be for the generic
target else the CUDA target cannot "see" it.
Further, as gpuci
is now passing (it wasn't in #8532, or by induction the merge to main
in 9d034e9) this also suggests the fix is working.
Make ol_compatible_view accessible on all targets
Buildfarm ID: Passed. |
As identified in numba#8271, the CUDA target needs to be set as the target at the bottom of the call stack, otherwise overloads for the generic target cannot be resolved. This is required so that the fix applied in numba#8562 (using the generic target for `ol_compatible_view` from numba#8537) actually works.
This is to fix Issue #8529, where test_reinterpret_array_type fails on CUDA with NumPy 1.23 because this overload is needed. Making it accessible to the CUDA target should resolve the issue.
Fixes #8529.