That’s because you are running it on hardware implementing IEEE 754 floating point.