PIC Microcontoller Math Method

Multiply 16x8 bit from Dingbat

;--------------------------------------------------------------------+
;   Function: Math : Multiply 16bits by 8bits                        |
;--------------------------------------------------------------------|
;   Filename: M_MUL16x8.inc       |   Environment    : MPLAB         |
;   Version : 1                   |   Microcontroller: PIC16F8x*     |
;   Date    : 21/07/2003          |   OSC MHz/MIPS   : xx / xxx      |
;--------------------------------------------------------------------|
;   Author         : Dingbat                                         |
;   Company        : -                                               |
;   Files required : none                                            |
;--------------------------------------------------------------------|

; Unsigned 16*8 bit Multiply AA(16) * BB(8), Result (CC) is 24bit
; adapted from code found on www.piclist.com
;
; params 	
;	AA 0:1 (MSB:LSB) 	= number to multiply (16bit)
;	BB					= multiplier (8bit)
;   CC 0:3 (MSB:LSB)	= Product (24bit)
;
; Taking out the loops and inlining the code would save about 
; another 20-30 instructions

MUL16x8
	clrf	CC+0		;clear result
	clrf	CC+1		;
	clrf	CC+2		;
	movlw	.8		
	movwf	bitCount	;bit counter set @ number of bits in multiplier

mul_L1					;main loop here
	rlf		CC+2,f			;rotate result left (*2)
	rlf		CC+1,f			;
	rlf		CC+0,f			;
	bcf		CC+2,0			;clear LSB of result after rotation
	btfss	BB,7			;test msb of multipier
	goto	dontAdd			;if bit is clear then dont add
	movfw   AA+1      		;Add the 16 bits of AA to product
	addwf   CC+2,f      	;
	skpnc					;
	incf	CC+1,f			;
	movfw   AA				;
	addwf   CC+1,f      	;
	skpnc					;
	incf	CC+0,f			;
dontAdd
	rlf		BB,f			;rotate multiplier left
	decfsz	bitCount,f		;chk if finished all bits in multiplier
	goto	mul_L1

	return

Comments:

• The problem is worse than the previous comment suggested. Only AA+1 (the LSByte) has to be zero for the routine to fail, which is a much bigger problem than if the whole AA register were zero+
• The modification suggested by Jordi Agust causes the multiply routine to fail if the AA register set is zero. The zero flag would only be valid if the increment above it was used. In the case of AA being zero, the zero flag is still set from the previous line "addwf CC+2,f", meaning CC+0 will get incremented, returning a non-zero result.+
• Someone said:

  I think it must be
  
  skpnz
  incf CC+0,f
  
  after the
  
  skpnc
  incf CC+1,f
  
  because the instruction incf does not affect the C flag (only the Z flag), so when you make the incf CC+1,f the carry flag doesn't become 0 when CC+1 isn't 00000000. But un CC+1 become 00000000 then the Z flag will become 1 so now I can detect the carry from CC+1 and increment CC+0.
  
  

• After the code
  
  skpnc
  incf CC+1,f
  
  should continue
  
  skpnc
  incf CC+0,f
  
  because the addition may be FF+1... and overflow the register.
  
  David Salgado+

  .