The _ _SUBT_C intrinsic has the following declaration:
double __SUBT_C(double, double);
This intrinsic function returns the result of subtracting the second argument from the first argument, with IEEE chopped rounding used. Both arguments and the result are in IEEE double-precision format.