Quick search


version 0.042

Python interface to GPU-powered libraries

People: Lev Givon


This SciKit (toolkit for SciPy) provides Python interfaces to a subset of the functions in the CUDA, CUDART, CUBLAS, and CUFFT libraries distributed as part of NVIDIA's CUDA Programming Toolkit, as well as interfaces to select functions in the free and standard versions of the CULA Dense Toolkit. In contrast to most existing Python wrappers for these libraries (many of which only provide a low-level interface to the actual library functions), this package uses PyCUDA to provide high-level functions comparable to those in the NumPy package.



You can download the latest distribution from PyPI here:

Easy Install

Install the Easy Install tools. Afterwards you can install scikits.cuda from the terminal by executing:

sudo easy_install scikits.cuda

If you prefer to do a local installation, specify an installation prefix:

easy_install --prefix=${HOME} scikits.cuda
and ensure that your PYTHONPATH is up to date, e.g.:
export PYTHONPATH=$PYTHONPATH:${HOME}/lib/python2.5/site-packages

This package was discovered in PyPI.