I am searching for a C++ library for arbitrary precision fixed-point arithmetic. It should be header-only and ideally have dynamic precision (so changeable precision at runtime). I searched the web for ages and could not find anything. I only need addition, subtraction and multiplication - so nothing too fancy. The library should be as performant as possible (I could also work with floating-point numbers, but fixed-point should be a lot faster, shouldn't it?). I have Boosts multiprecision floats working, but they can get really slow. I gave up on getting GMP to work (Windows) - also on that Windows-fork of it who's name I can't remember...
Does anybody have any recommendations?
