On Fri, Jul 12, 2002 at 03:42:43PM -0300, Dennis Crawley wrote:
> Hi all,
> I need suggestions on these Algorithms, or the best way to get:
> 1.DATAIN contains the bit(position) to be set in DATAOUT.
> 2.DATAOUT have to be changed without change any
> other bit than the position given by DATAIN.
>
>     Ideal: BCF    DATAOUT,[VALUE IN DATAIN]

Which of course doesn't exists.

>
> Algorithm 1: 7 instr. 7 to 35 cycles
>  MOVLW  0X05  ; Dummy: bit to be changed (1 to 8, not 0 to 7)
>  MOVWF  DATAIN  ; (used as a decrement counter)
>  BSF  STATUS, C ; assuming temp=0 at init.
> SetBit
>  RLF  temp,F   ;
>  DECFSZ DATAIN,F ;
>  GOTO SetBit   ;
>  MOVFW temp   ;w=temp
>  IORWF DATAOUT,F  ;set bit 5(4) living all bits untouched
>
> Algorithm 2: 14 instr. 12 cycles

You need a setup instruction
   MOVF  DATAIN,W

>  LCALL BITSELEC
>  CLRF PCLATH
>  MOVWF DATAOUT

The above instruction needs to be:
   IORWF DATAOUT,F


>
> BITSELEC
>  ANDLW 0X0F

And this instruction above needs to be:

   ANDLW 0x07

>  ADDWF PCL,F
>  RETLW 0X00
>  RETLW 0X01
>  RETLW 0X02
>  RETLW 0X04
>  RETLW 0X08
>  RETLW 0X10
>  RETLW 0X20
>  RETLW 0X40
>  RETLW 0X80

Algorithm 2 is definitely along the right lines. It's better because it
takes a constant amount of time to execute no matter which bit is being set.

But you're on the right track. In my bytecode interpreter I used the the
following fragment which is analogous to Algorithm 2

   BCF   STATUS,C
   RLF   DATAIN,W
   ANDLW 0x0F
   ADDWF PCL,F

   BSF   INDF,0
   RETURN
   BSF   INDF,1
   RETURN

   ; ...

But it's the same idea as your algorithm 2.

BAJ

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics