Split a double-precision floating-point number into a higher order word and a lower order word.
Create a double-precision floating-point number from a higher order word and a lower order word.
Set the less significant 32 bits of a double-precision floating-point number.