death to floating point

@novenary So fixed point numbers?

@vavency yes, and fractions with carefully chosen denominators, eg 120 which is highly composite

@novenary Back to 16bits per point side.

@vavency you can decide how much precision you need which is better than wasting a quarter of your bits on the exponent + you avoid all the other problems floats have

@novenary @vavency fixed point operations are also for the most part exactly the same as integer operations