There are several resources like: http://codebase64.org/doku.php?id=base:kernal_floating_point_mathematics but because of lack of example I cant use Floating Point routines in Kernal.

I need to divide, multiply 16 bit values, for example: $1234*$ABCD or $1234/$ABCD

I assume that I must be: 
1. Convert 16 values into FAC value, 
2. Then call division or multiply operation, 
3. And convert FAC to 16 bit value again, 

But How?