Sigh... Yet another one.  I guess I'll have to remove the
    "Reply-To" header from my outgoing e-mail.

    -Andy

------- Forwarded Message Follows -------

From:          John Payson <supercat@Mcs.Net>
Subject:       Re: (Fwd) Re: (Fwd) Re: Bit Number to Bit Mask Question
To:            fastfwd@ix.netcom.com
Date:          Wed, 17 Sep 1997 20:53:18 -0500 (CDT)


> From:          Dmitry Kiryashov <zews@aha.ru>
>
> ;       Minimizing ... Final pass .
> ;
>  movlw  $01
>  btfsc  BITNUM,1
>  movlw  $04
>  movwf  temp
>
>  btfsc  BITNUM,0
>  addwf  temp,f

Ah... good way to do the conditional shift without worrying about
carry flag. Cool.
>
>  btfsc  BITNUM,2
>  swapf  temp,f
>
>  movfw  temp
>
> ;       9 words / 9 cycles

That's a savings.  Probably makes it as good as the table for many
app- lications.

> >  movlw  $FF
> >  btfsc  BITNUM,0
> >   andlw $AA
> >  btfsc  BITNUM,1
> >   andlw $CC
> >  btfsc  BITNUM,2
> >   andlw $F0
> >  movwf  temp
>
> ;       Optimizing ...
>
>  sublw  0
>  andwf  temp,w
> ;
> ;       10 words / 10 clocks

Okay, the SUBLW seems to have been the trick I was looking for.

> > Interesting, but even worse (unless I'm missing an optimization on
> > that ending part).  Anyone see any improvements for those two methods
> > above?

> The first code looks better .

I guess, it does seem to be a word shorter/cycle faster.  Though
there's something I like aesthetically about the second... I'll just
have to find a way to rid of those 4 annoying cycles (and the use of
TEMP)...

------- End of Forwarded Message -------