fix: ops_summary shows GPU kernel names for orphan HIP launches (#641)#642
Open
gabeweisz wants to merge 4 commits into
Open
fix: ops_summary shows GPU kernel names for orphan HIP launches (#641)#642gabeweisz wants to merge 4 commits into
gabeweisz wants to merge 4 commits into
Conversation
When no cpu_op ancestor exists, the launcher row is cuda_runtime (hipLaunchKernel). ops_summary grouped by that name, merging distinct custom kernels. Substitute kernel name(s) from kernel_details for cuda_runtime/cuda_driver launchers, matching unified perf behavior. Fixes #641 Co-authored-by: Cursor <cursoragent@cursor.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes #641.
When GPU work is launched from a standalone
hipLaunchKernel(or other cuda_runtime/cuda_driver) row with nocpu_opancestor,ops_summarygrouped every such launch under the same name. The DataFrame now uses the actual kernel name(s) fromkernel_details, consistent with unified perf synthetic naming.Changes
TreePerfAnalyzer._kernel_launcher_ops_summary_name: forcuda_runtime/cuda_driverlaunchers withkernel_details, set the rownameto the kernel name (comma-separated if multiple).ops_summaryrows.Testing
pytest tests/test_kernel_launchers.py(21 passed).Made with Cursor