Parallel CPU, GPU processing - benchmark, bandwidth, bottleneck and limitation. The code has been tested on Python 3.4.3 x64, CUDA 7.0 x64 and Microsoft Windows 10 x64. Here are some of the extra Python packages which are needed to run this code:
- Numpy
- Pycuda
- Visual C++ compiler and linker
Documentation:
http://hubonit.com/ideas/?p=671