Samuli Laine

Sorry it took a while but here goes. The computations are done using [fixed-point arithmetic](https://en.wikipedia.org/wiki/Fixed-point_arithmetic), as the CPU doesn't support native floating-point calculations and emulation of those would be slow....