[Uberphallus]

Looks like a complex topic, but the code is small enough to help remove the rust from my math gears. I guess everybody would want it ported to CUDA/OpenCL/Vulkan?