We solved that for floats like 10 years ago. Let alone the fact there are better formats, like posits, or fixed point numbers, that also solve this problem very easily.