README

IMPORTANT PUBLIC SERVICE ANNOUNCEMENT:

This has been developed and tested under the CUDA 5.0 toolkit. I am not
testing against CUDA 5.5.

Reasons: CUDA 5.0 is the lowest version to fully support the full range
         of nVidia GPUs on the market!

         CUDA 5.0 has lower minimum driver requirements than CUDA 5.5
         and is hence compatible with a larger install base.

         CUDA 5.0 has all the features I require for this project.

Recent commits have shown to either not compile on CUDA 5.5 (PTXAS
throwing some error) or to not run properly on Kepler devices (e.g.
the K520 grid GPU). I have no time to debug or fix this (if it is
fixable at all).

Do yourself a favor and please get the CUDA 5.0 toolkit!

Otherwise...

See README.txt