> > As an exercise for the viewer, if you use SSE2, prefetch, and > non-branching bitmath you can perform this roughly 32 times as fast. > > Regards, > Daniel Please teach me!