But which advantage would this have over using IEEE 754-2008 decimal floating point numbers?