Open
Conversation
IvanGrigorik
force-pushed
the
grigorik/binding-fix
branch
from
fcd95d9 to
92213e3
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR fixes the binding issue for the device-sided views that are passed into host-sided kernels.
TLDR: if the default space is device, but we call the host kernel - there will be an issue with argument memory layout (Left vs Right). This case should be handled by our side
When a host-space kernel (e.g., pk.Serial) is run in a device context, the Python side passes views that live in GPU memory (CudaUVMSpace, LayoutLeft), while the generated C++ binding expects host views with the host execution space’s layout (HostSpace, LayoutRight for Serial).
Pybind11 can't automatically cast these types, so we should handle that on our side.
This PR introduces function
_generate_mirror_with_exec_layoutwhich is handling this caseThis PR also adds a compilation check. We should always check if the concrete module file exists, not the module directory.