I'm pretty new to gcc, so forgive me if this is a stupid question: is there any way to extract more info about a floating point type than just its size? In particular, I'd like to see how many bits the significand takes up. Ideally this should be done at run time, but that's not essential.