[UR][HIP] Enable usm pools #17972
Merged
[UR][HIP] Enable usm pools #17972
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
npmiller
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
The pool helper works on device pool and isn't suitable for host allocations.
npmiller
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
aarongreig
reviewed
Apr 11, 2025
igchor
reviewed
Apr 11, 2025
The helper was using the device pool for all of the allocations resulting in crashes.
igchor
approved these changes
Apr 11, 2025
frasercrmck
approved these changes
Apr 14, 2025
|
@intel/llvm-gatekeepers this is ready to merge |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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 patch fixes up and enable memory pools for the HIP adapter, it is based on oneapi-src/unified-runtime#1689 and on the CUDA adapter implementation.
The initial patch had segmentation faults in the CI that we couldn't reproduce locally. That happened as well in this patch and I couldn't reproduce the segfaults locally either.
However I noticed that it failed in
urUSMHostAlloc, and that entry point was different from the CUDA adapter version, in that the HIP adapter was using a "helper" function. It turns out that the helper function was using a device pool instead of a host pool to do the allocation, which seemed obviously wrong. Replacing the helper by similar code used in the CUDA adapter fixes the crash in the CI.