Fix GPU handle pool singleton aliasing #525
Merged
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
Arech8
reviewed
Aug 1, 2025
|
Are you sure the PR does what you describe in the title and description? |
phambinhfin
force-pushed
the
phambinh/handle-pool-singleton
branch
from
edd7ffb to
cf8b52f
Compare
phambinhfin
changed the title
phambinhfin
force-pushed
the
phambinh/handle-pool-singleton
branch
from
cf8b52f to
4d49336
Compare
|
@Arech8 I modified the content of commit and title, thanks |
Ruturaj4
approved these changes
Aug 1, 2025
Arech8
reviewed
Aug 1, 2025
jaxlib/gpu/vendor.h
Outdated
| return hipblasCreate(reinterpret_cast<hipblasHandle_t*>(handle)); | ||
| } | ||
|
|
||
| inline hipblasStatus_t gpublasSetStream(gpublasHandle_t handle, hipStream_t stream) { |
There was a problem hiding this comment.
Why do you need this one? Does it work if you leave it as define?
There was a problem hiding this comment.
Sorry for being a nitpick here. Does #define gpublasSetStream hipblasSetStream work? I was expecting that to work.
There was a problem hiding this comment.
Currently, i am set like
#define gpublasCreate(handle) hipblasCreate(reinterpret_cast<hipblasHandle_t*>(handle))
#define gpublasSetStream(handle, stream) hipblasSetStream(reinterpret_cast<hipblasHandle_t>(handle), stream)
It works,
phambinhfin
force-pushed
the
phambinh/handle-pool-singleton
branch
from
4d49336 to
2bc0153
Compare
…tion Problem: - BLAS and SOLVER handle pools were sharing the same static singleton instance - C++ template instantiation created only one pool when HandleType/StreamType were identical - This caused resource contamination between hipBLAS and hipSOLVER operations - Led to potential memory corruption and unexpected GPU library behavior
phambinhfin
force-pushed
the
phambinh/handle-pool-singleton
branch
from
2bc0153 to
f9ef066
Compare
zahiqbal
pushed a commit
that referenced
this pull request
Oct 8, 2025
Addresses handle pool singleton sharing issue between different GPU operation types in ROCm/HIP backend.
zahiqbal
pushed a commit
that referenced
this pull request
Oct 8, 2025
Addresses handle pool singleton sharing issue between different GPU operation types in ROCm/HIP backend. (cherry picked from commit 4b5b368)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Addresses handle pool singleton sharing issue between different GPU
operation types in ROCm/HIP backend.
Changes made: