Fix race condition in OpenCL kernel #3535
Open
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Outdated suggestions cannot be applied.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.
Description
Without the barrier at the end of barrierOR, it is possible for work-item 0 to start the next loop iteration and update predicates[0] while other work-items are still inside barrierOR reading
predicates
, meaning they read the next loop iteration's exit condition. This results in a divergent loop, where not all work-items reach the same barriers.A previous fix identified this as a problem only on NVIDIA platforms, but strictly speaking a barrier is required in all cases to avoid a spec violation and undefined behaviour.
Changes to Users
The kernel should produce correct results on more OpenCL implementations.
Locally I tested both
Intel(R) FPGA Emulation Device
and various oneAPI Construction Kit devices, which all previously failed theconfidence_connected_opencl --gtest_filter="SingleSeed/ConfidenceConnectedDataTest.SegmentARegion/_prefix_background_radius_0_multiplier_1_iterations_5_replace_255"
unit test.I'm unable to test other OpenCL implementations, sorry.
Checklist
[ ] Functions added to unified API[ ] Functions documented