fix GPU wheel: patch AMReX C++ sources for CUDA 12 NVTX header#260
Merged
jameslehoux merged 1 commit intomasterfrom Apr 21, 2026
Merged
fix GPU wheel: patch AMReX C++ sources for CUDA 12 NVTX header#260jameslehoux merged 1 commit intomasterfrom
jameslehoux merged 1 commit intomasterfrom
Conversation
Previous fix rewrote CUDA::nvToolsExt -> CUDA::nvtx3 in the AMReX CMake
target, but AMReX 25.03 also has a bare
# include <nvToolsExt.h>
in Src/Base/AMReX_GpuDevice.cpp (and a few other places). CUDA 12 ships
the header-only NVTX3 replacement at <nvtx3/nvToolsExt.h> instead, so
the bare include fails with
fatal error: nvToolsExt.h: No such file or directory
Add a second sed pass over /tmp/amrex/Src that rewrites every
#include <nvToolsExt.h>
to
#include <nvtx3/nvToolsExt.h>
regardless of whitespace between `#` and `include`. The anchor `<` in
the pattern prevents double-prefixing files that already use the nvtx3/
path. Bumped the cache key to v5 to force a clean AMReX rebuild.
Verified locally: mixed whitespace forms get rewritten, files already
using <nvtx3/nvToolsExt.h> are untouched.
https://claude.ai/code/session_011dJ5Bwq4Tnr8wxH597XJFf
Code Coverage ReportGenerated by CI — coverage data from gcovr |
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Previous fix rewrote CUDA::nvToolsExt -> CUDA::nvtx3 in the AMReX CMake target, but AMReX 25.03 also has a bare
in Src/Base/AMReX_GpuDevice.cpp (and a few other places). CUDA 12 ships the header-only NVTX3 replacement at <nvtx3/nvToolsExt.h> instead, so the bare include fails with
Add a second sed pass over /tmp/amrex/Src that rewrites every
to
regardless of whitespace between
#andinclude. The anchor<in the pattern prevents double-prefixing files that already use the nvtx3/ path. Bumped the cache key to v5 to force a clean AMReX rebuild.Verified locally: mixed whitespace forms get rewritten, files already using <nvtx3/nvToolsExt.h> are untouched.