functions return the larger of
and likewise, the
functions return the smaller of
as being larger than
If one argument is an NaN, then the other argument is returned.
If both arguments are NaNs, then the result is an NaN.
These routines do not raise any floating-point exceptions.