Release GIL when calling mkl_lapack::orgqr#2850
Merged
antonwolfy merged 4 commits intomasterfrom Apr 21, 2026
Merged
Conversation
Contributor
|
View rendered docs @ https://intelpython.github.io/dpnp/index.html |
Contributor
|
Array API standard conformance tests for dpnp=0.20.0dev6=py313h509198e_13 ran successfully. |
Collaborator
|
coverage: 78.429%. remained the same — add-gil-release-to-orgqr into master |
antonwolfy
requested review from
ndgrigorian and
vlad-perevezentsev
as code owners
Co-authored-by: vlad-perevezentsev <vladislav.perevezentsev@intel.com>
vlad-perevezentsev
approved these changes
Apr 14, 2026
Contributor
vlad-perevezentsev
left a comment
vlad-perevezentsev
left a comment
There was a problem hiding this comment.
LGTM
Thank you @antonwolfy
github-actions bot
added a commit
that referenced
this pull request
Apr 21, 2026
This PR fixes a deadlock in QR decomposition tests by adding GIL release
before `mkl_lapack::orgqr` call.
The hang occurred because:
1. dpnp/dpctl submits a `host_task` to manage Python object lifetimes
2. host_task needs to acquire the GIL to decrement reference counts
3. if the main thread holds the GIL during queue submission → deadlock
4. `orgqr` is currently implemented in oneMKL as GPU-to-Host reverse
offload:
```cpp
exec_q.submit([&](sycl::handler& cgh) {
cgh.depends_on(depends);
cgh.host_task([=]() { orgqr_host(...); });
}).wait();
```
As a solution PR proposes to release GIL using before calling the OneMKL
operations. The GIL is automatically reacquired when the function
returns (RAII). 3ab1216
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This PR fixes a deadlock in QR decomposition tests by adding GIL release before
mkl_lapack::orgqrcall.The hang occurred because:
host_taskto manage Python object lifetimesorgqris currently implemented in oneMKL as GPU-to-Host reverse offload:exec_q.submit([&](sycl::handler& cgh) { cgh.depends_on(depends); cgh.host_task([=]() { orgqr_host(...); }); }).wait();As a solution PR proposes to release GIL using before calling the OneMKL operations. The GIL is automatically reacquired when the function returns (RAII).