Floating point arithmetic doesn't «suck», it's the code that treats floats as decimal numbers that sucks.