fix LAMMPS wheel with CUDA wheels #2887
Merged
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Outdated suggestions cannot be applied.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.
It's annoying that in the CUDA Python wheels, the library filename is
libcudart.so.11.0
, instead oflibcudart.so
orlibcudart.so.11
orlibcudart.11.0.xxx
. There are no symlinks in these CUDA Python wheels on PyPI. Thus, in this case, when DeePMD-kit tries to loadlibcudart.so
, no file can be found.Python interface doesn't have this issue, as it tries to use
glob
to find detailed library names. For the LAMMPS executable called from Python,LD_LIBRARY_PATH
doesn't work with CUDA Python libraries (but works with an installation that contains symlinks). Thus, in this PR,LD_PRELOAD
is set instead to preload CUDA symbols.This fixes the docker images for the development branch.