Addition:
Carries:              1 11111111 1111111
> Number 1:    00000000 11111111 11111111
> Number 2:    01100100 11111111 11111111
               --------------------------
Answer:        01100101 11111111 11111110

> The simulator gives out the 24 bit result:
>              01100110 00000010 11111101

Looks like some problem related to handling carries between bytes, but
since the error isn't consistent I can't guess what exactly is wrong.

Multiplication:
> Number 1:  00000000 00000010
> Number 2:  00000000 00000001
>
> And get the output multiplied
>            00010000 00010000

The correct answer is, of course, 2 (00000010).  The routine you found
either doesn't do what you think it does, was typed in incorrectly, or
perhaps you didn't properly assign the input/output/temp registers it
needs.

I hope you weren't expecting any more specific answers, given that you
didn't post the code that produced these strange results.
        Jason Harper