Could we implement IEEE quadruple-precision (binary128) format?