In my endless pursuit of near-perfection again. That was the last place i was expecting there to be bottlenecks. The bad news is that I really can't make this function any simpler aside from maybe skipping rotation when 0. The good news is that GCC is having no trouble at all optimizing this part and it's unlikely this will seriously need to be addressed before release, so long as I'm careful. The parts of the code I've been careful to simplify have barely even registered in the benchmark which means I am doing an excellent job so far. I have been documenting the best practices for performance in the MYTHOS manual as I determine them.