The
fdim(),
fdimf(), and
fdiml() functions return the positive difference between
x and
y. That is, if
x-y is positive, then
x-y is returned. If either
x or
y is an NaN, then an NaN is returned. Otherwise, the result is
+0.0.
Overflow or underflow may occur iff the exact result is not representable in the return type. No other exceptions are raised.