I believe the encoding for "Floating-point<->integer conversions", and "Floating-point<->fixed-point conversions" is completely incorrect.