How the bits of a floating point number are organized, how (de)normalization works, etc.| John D. Cook