Search code examples
numpycythonctagsmex

Wrapping C++ and CUDA in Python

I want to create an interface for a numerical library consisting of both OOP C++ (boost) and CUDA C code, in Python. There is already an existing MATLAB interface, but it contains a lot of mex.h dependencies.

How can this be done as painless as possible?

Solution

  • Here are a couple of links to look at. Could people who've used any of these please comment ?

    # day  status  packagename  version  homepage  summary

2011-02-03  4  "scikits.cuda"  0.03  http://github.com/lebedov/scikits.cuda/
    Python interface to GPU-powered libraries

2010-10-27  0  "KappaCUDA"  1.5.0  http://psilambda.com
    Module to give easy access to NVIDIA CUDA from Python using the Kappa Library.

2010-10-16  5  "pycuda"  0.94.2  http://mathema.tician.de/software/pycuda
    Python wrapper for Nvidia CUDA

2010-07-01  4  "PyGouda"  1.0  http://pypi.python.org/pypi/pycuda
    The EasyCheese of GPU programming