Regulus Berdin wrote:
>
> >
> > Close, but try X = 0x95 and Y = 0x95. The correct answer should
> > be Y=0x80 and C=1. However, this routine will produce Y=0x2A and
> > C=1. You still need that annoying normalization step.
>
> This code will produce 90 with C=1.
> Should be the right answer 95+95 = 190 not _180_.
> The code must be correct!

Uhh, yeah. Sorry. But what about this case: 95 + 90

BCD_ADD:                        ;Y = Y + X
        MOVLW   66               W = 66
        ADDWF   X,W              W = 95 + 66 = FB
        ADDWF   Y,F              Y = 90 + FB = 8B, C=1, DC=0
        MOVLW   -60              W = -60 = 0xa0
        SKPDC                    NOT SKIPPED
         MOVLW  -66               W = -66 = 0x9a
        SKPC                     SKIPPED
         ADDWF  Y,F               NO ADD, SO RESULT = 8B, C=1
        RETLW   0

If C=1 AND the sum of the least significant nibbles is less than 10
then the result will be incorrect.


>
> >
> > > BCD_MINUS:                      ;Y = Y - X
> > >         MOVF    X,W
> > >         SUBWF   Y,F
> > >         MOVLW -6
> > >         SKPC
> > >          MOVLW  -66
> > >         SKPDC
> > >          ADDWF  Y,F
> > >         RETLW   0
> >
> > Try X=0x25, Y=0x16. This produces 0xf1 instead of 0x91.
> I goofed on this one :(.
>
> How about this:
> BCD_MINUS:              ;Y = Y - X
>         MOVF    X,W
>         SUBWF   Y,F
>         ANDLW   0       ;Set Zero flag
>         SKPDC
>          IORLW  6
>         SKPC
>          IORLW  60
>         SUBWF   Y,F     ;Zero flag is cleared
>         RETLW   0
> Test ZERO flag for underflow.

Excellent! However from my simulation results, Z is set only when
the result is 0 (and not for an underflow). If you macro-ize this
routine then the upper nibble of W will have the carry information.
Or perhaps, you can add :

        SUBWF   Y,F
        ANDLW    0x60   ;Set Z if underflow occurred
        RETLW    0


Scott