So can you implement soft floating point math in soft integer math in floating point math?