Safe manipulating with large integers (rather than 32bits)
Return an unsigned 32-bit integer corresponding to the more significant 32 bits of a double-precision floating-point number.
Set the more significant 32 bits of a double-precision floating-point number.