[ROCm] Enable Triton MLIR verifier by default on ROCm#786
Open
nurmukhametov wants to merge 1 commit intomainfrom
Open
[ROCm] Enable Triton MLIR verifier by default on ROCm#786nurmukhametov wants to merge 1 commit intomainfrom
nurmukhametov wants to merge 1 commit intomainfrom
Conversation
Add a new flag xla_gpu_rocm_triton_verifier (default true) that enables the MLIR pass verifier in the Triton compilation pipeline on ROCm. This turns invalid IR from miscompiling passes into graceful compilation errors that the autotuner can skip, instead of crashing the process.
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
Review SummaryClean, low-risk change that enables the MLIR pass verifier by default on ROCm to catch invalid IR early. Code logic is correct — the OR condition, flag registration, and default value are all properly wired up. Two minor documentation nits left inline:
No functional concerns. 🤖 Generated with Claude Code |
github-actions
bot
removed
the
claude-review
i-chaochen
requested review from
amd-jianli12,
mfrancepillois and
zoranjovanovic-ns
Comment on lines
+2946
to
+2953
| flag_list->push_back(tsl::Flag( | ||
| "xla_gpu_rocm_triton_verifier", | ||
| bool_setter_for(&DebugOptions::set_xla_gpu_rocm_triton_verifier), | ||
| debug_options->xla_gpu_rocm_triton_verifier(), | ||
| "If true, enables MLIR pass verification in the Triton compilation " | ||
| "pipeline on ROCm. Catches invalid IR early so the autotuner can skip " | ||
| "bad configs instead of crashing. Only consulted when " | ||
| "xla_gpu_llvm_verification_level is 0 and the target is ROCm.")); |
There was a problem hiding this comment.
Why do we need a new flag for this? Couldn’t we just use the existing xla_gpu_llvm_verification_level flag? (Perhaps we could add another level (e.g. 2) if you don’t want to enable the verifier in emitter_base.cc).
Member
Author
There was a problem hiding this comment.
I was concerned to anyhow change xla_gpu_llvm_verification_level flag's behaviour especially bringing to it ROCm specific one.
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.
Add a new flag xla_gpu_rocm_triton_verifier (default true) that enables the MLIR pass verifier in the Triton compilation pipeline on ROCm. This turns invalid IR from miscompiling passes into graceful compilation errors that the autotuner can skip, instead of crashing the process.
Submission Checklist