Important
The final release of this project was v0.7.0.
Similar functionality is available in cuda.core, and for numba-cuda>=0.16,
numba-cuda automatically detects and enables nvjitlink when needed and available with no explicit configuration.
See https://docs.rapids.ai/notices/rsn0052/ for details.
The RAPIDS pynvjitlink library provides a Python binding for the nvJitLink library.
pip install pynvjitlink-cu12Install with either:
python -m pip install .or
python -m pip install -e .for an editable install.
conda install -c rapidsai pynvjitlinkReview the CONTRIBUTING.md file for information on how to contribute code and issues to the project.