[dahart]

I am talking about float operations, of course. And they’re all inaccurate, generally speaking, because they round. Fast math rounding error is not much larger than rounding error without fast mast.