These functions round their argument to the nearest integer value, rounding away
from zero, regardless of the current rounding direction. If x is
infinite or NaN, or if the rounded value is outside the range of the return
type, the numeric result is unspecified. A domain error may occur if the
magnitude of x is too large.