OH GOD THE CODE IT'S UGLY. (C++ changes)

[Bitweasil]

Fortunately, I'm fixing it.

Working on a rewrite of the multiforcer to C++, multi-GPU, and more extensible. The old code, if anyone looked at it, was truly hideous. I've also found some places I wasn't effectively using the multithreaded nature of GPU kernels to coalesce memory accesses, so... hopefully there will be some minor speedups there.