[release/v0.1.12] Remove C10_HIP_KERNEL_LAUNCH_CHECK (Lingpeng's preferred fix)#2846
Merged
valarLip merged 1 commit intorelease/v0.1.12from Apr 22, 2026
Merged
Conversation
…nels.cu Per Lingpeng's review on PR #2839/#2842 — the macro is defined only by recent PyTorch versions' c10/hip/HIPException.h. Building AITER from source against older PyTorch releases (e.g. vLLM ml-ci-internal lane, issue #2837) fails to find the macro even when the include is present. The macro is used in exactly one place across all of csrc/. It's an error-reporting wrapper around hipGetLastError() — the kernel launches themselves are unaffected. Removing it eliminates the build-env dependency and is robust across PyTorch versions. Supersedes #2842 (cherry-pick of #2839's include-based fix). Fixes #2837.
ganyi1996ppo
approved these changes
Apr 22, 2026
This was referenced Apr 22, 2026
Merged
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Remove the
C10_HIP_KERNEL_LAUNCH_CHECK();call fromcsrc/kernels/gated_rmsnorm_quant_kernels.cu(the only occurrence across all ofcsrc/). This is Lingpeng's preferred fix per Teams discussion 2026-04-21 and supersedes the include-based approach in #2842.Background
ml-ci-internallane fails to build AITER becauseC10_HIP_KERNEL_LAUNCH_CHECKis undeclared.#include <c10/hip/HIPException.h>.c10/hip/HIPException.h. The include-based fix is brittle across PyTorch versions.Why removal is safe
C10_HIP_KERNEL_LAUNCH_CHECKis a thin error-reporting wrapper aroundhipGetLastError(). The kernel launches themselves (<<<grid, block, 0, stream>>>) are unaffected. Dropping the post-launch check removes the build-environment dependency on PyTorch's HIPException header and makes the file robust across PyTorch versions.The macro is used in exactly one place across
csrc/(verified). Onrelease/v0.1.12HEAD (04d4009) the include from #2842 is not yet present, so no include needs to be removed.Relationship to other PRs / issues
CI
Adding labels for vLLM, ATOM, SGLang, and Triton MI355 lanes per release/v0.1.12 validation policy.