Use GPU acceleration

Tasks like this would be well suited for GPU acceleration, and I'm sure both ATI and Nvidia would be happy to give some help when it comes to using CUDA and Streams. this would give your product a serious speed advantage over the competition, while costing nothing to those that don't have GPUs capable (as it would be the same speed as before). Also 64bit support would be nice, as the wider register pointers would also help speed things up.