Do not try to use i8 and i16 versions of FP_TO_U/SINT soft float library calls