[GPU][PTL] WA: Set max WG size to 512 for reduce_ref kernel in PTL#35072
Merged
isanghao merged 1 commit intoopenvinotoolkit:masterfrom Apr 2, 2026
Merged
Conversation
isanghao
reviewed
Apr 1, 2026
| dispatchData.lws = GetOptimalLocalWorkGroupSizes(dispatchData.gws, engineInfo, in_layout, out_layout, dims_by_gws); | ||
| } else { | ||
| dispatchData.lws = GetOptimalLocalWorkGroupSizes(dispatchData.gws, params.engineInfo, in_layout, out_layout, dims_by_gws); | ||
| } |
Contributor
There was a problem hiding this comment.
what about just limiting to 512 unconditionally? as this is a ref kernel, I hope it should be fine.
Contributor
Author
There was a problem hiding this comment.
This is just a workaround code. Since it will be removed once the driver issue is fixed, I tried to keep the changes as simple as possible. Is it necessary to add a new function or logic here?
| try { | ||
| _command_queue.enqueueNDRangeKernel(kern, cl::NullRange, global, local, dep_events_ptr, set_output_event ? &ret_ev : nullptr); | ||
| } catch (cl::Error const& err) { | ||
| if (err.err() == CL_INVALID_WORK_GROUP_SIZE && local != cl::NullRange) { |
Contributor
There was a problem hiding this comment.
not sure whether it is a proper place to handle this failure. We have other error code, but why should it be handled differently?
Contributor
Author
There was a problem hiding this comment.
It will be removed after fixing the driver issue. If this debugging code doesn’t seem appropriate, I’ll remove it.
3 tasks
9f99a0f to
8d24ed8
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description of the issue (symptom, root-cause, how it was resolved)
Reduce kernel CL_INVALID_WORK_GROUP_SIZE on PTL (-54)
CL_INVALID_WORK_GROUP_SIZE(-54) on Panther Lake (xe3 integrated GPU)maxWorkGroupSizeto 512 before computing optimal LWS inreduce_kernel_ref.cppThe code and line that caused this issue (if it is not changed directly)
intel_gpu/src/kernel_selector/kernels/reduce/reduce_kernel_ref.cppChecklist
Tickets: