Normalize CUDA_ARCH_LIST and add explicit NVCC gencode flags for >SM99 #123
Conversation
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
- honor CUDA_ARCH_LIST/CT2_CUDA_ARCH_LIST in build.rs and normalize tokens - keep legacy architectures in CUDA_ARCH_LIST for CMake, but emit -gencode flags for SM major >= 10 so new GPUs (Ada/Blackwell) bypass FindCUDA’s limits - allow optional extra NVCC flags to coexist with the small-binary compression flag
Owner
|
Thanks for the PR! I tested it locally and it works great. Merging now. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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Outdated suggestions cannot be applied.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
CMake’s built-in
select_compute_arch.cmakehas no notion of the SM 8.9/12.0targets used by Ada and Blackwell GPUs, so passing those numbers through
CUDA_ARCH_LIST stops the build. This PR updates ct2rs’s build script to
normalize the env var tokens, keep older arches in CUDA_ARCH_LIST, and add
explicit
-gencodeflags for anything with a major version ≥ 10. That letsnewer GPUs compile cleanly without regressing older cards.
Usage
CUDA_ARCH_LIST(e.g.export CUDA_ARCH_LIST="8.9;12.0").cargo build --release --features "cuda,ct2rs/cuda,ct2rs/cudnn".newer ones automatically get direct NVCC
-gencodeentries.