Pyculib

Pyculib provides Python bindings to the following CUDA libraries:

These bindings are direct ports of those available in Anaconda Accelerate.

Documentation is located here

Installing

The easiest way to install Pyculib and get updates is by using the Anaconda Distribution

#> conda install pyculib

To compile from source, it is recommended to create a conda environment containing the following:

for instructions on how to do this see the conda documentation, specifically the section on managing environments.

Once a suitable environment is activated, installation achieved simply by running:

#> python setup.py install

and the installation can be tested with:

#> ./runtests.py

Documentation is located here.

It is also possible to build a local copy of the documentation from source. This requires GNU Make and sphinx (available via conda).

Documentation is stored in the doc folder, and should be built with:

#> make SPHINXOPTS=-Wn clean html

This ensures that the documentation renders without errors. If errors occur, they can all be seen at once by building with:

#> make SPHINXOPTS=-n clean html

However, these errors should all be fixed so that building with -Wn is possible prior to merging any documentation changes or updates.