Installation#

Requirements#

CUDA Python is supported on all platforms that CUDA is supported. Specific dependencies are as follows:

  • Driver: Linux (450.80.02 or later) Windows(456.38 or later)

  • CUDA Toolkit 11.0 to 11.7

Note

Only the NVRTC redistributable component is required from the CUDA Toolkit. CUDA Toolkit Documentation Installation Guides can be used for guidance. Note that the NVRTC component in the Toolkit can be obtained via PiPy, Conda or Local Installer.

Installing from PyPI#

pip install cuda-python

Installing from Conda#

conda install -c nvidia cuda-python

Installing from Source#

Requirements#

Building dependencies:

  • cython>=0.29.24

Unit-test dependencies:

  • pytest>=6.2.4

  • pytest-benchmark>=3.4.1

  • numpy>=1.21.1

Latest list of dependencies can be found at requirements.txt

As per release 11.7.1, the latest CTK headers are required. Mobile platforms facing “has not been declared” error may use headers in https://gitlab.com/nvidia/headers/cuda as a temporary workaround. Consult the release notes for further direction.

In-place#

To compile the extension in-place, run:

python setup.py build_ext --inplace

To compile for debugging the extension modules with gdb, pass the --debug argument to setup.py.

Develop#

You can use

pip install -e .

to install the module as editible in your current Python environment (e.g. for testing of porting other libraries to use the binding).

Build the Docs#

conda env create -f docs_src/environment-docs.yml
conda activate cuda-python-docs

Then compile and install cuda-python following the steps above.

cd docs_src
make html
open build/html/index.html

Publish the Docs#

git checkout gh-pages
cd docs_src
make html
cp -a build/html/. ../docs/