On Tue, 10 Nov 2009, Michel Bouissou wrote:
> Hi there,
>
> Well, I've finally converted my PIC16 project to PIC18, and I was taking a
> look
> at the first SDCC .asm output that accepted to compile OK.
>
> My attention was drawn by the following SDCC-generated comment in the .asm
> file
> :
>
> ; ;multiply lit val:0x02 by variable _resetSource and store in r0x00
> ; ;Unrolled 8 X 8 multiplication
> ; ;FIXME: the function does not support result==WREG
>
> This pops 2 questions in my mind :
>
> 1/ Is this "FIXME" the sign of anything I should worry about ? ;-)
>
> 2/ As the PIC18 datasheet says "All PIC18 include an 8x8 hardware multiplier
> as part of the ALU [...] Making multiplication a hardware operation allows it
> to be completed in a single instruction cycle".
>
> And this corresponds to the PIC18 "MULLW" and "MULWF" asm instructions.
>
> Datasheet-provided "performance comparison" chart shows that a 8x8 unsigned
> MULT will take 1 CPU cycle and use 1 program memory word using the hardware
> multiplier, where it would use 69 (max) CPU cycles and use 13 program memory
> words if performed "the software way" in assembly code.
>
> And the SDCC .asm file when I saw the comment that drew my attention contains
> no trace of ever using a MULLW or MULWF, so I assume it performs
> multiplications the long, soft way...
>
> Any light on these 2 questions would be greatly appreciated...
Saw this posting recently which made me think... Especially as I'm doing a
lot of floating pint work which would no-doubt be accellerated by the
hardware multiply, however I've just had a look at some of the code in my
project... Although it has that warning, it is using the MUL instructions.
E.G.:
; ;multiply lit val:0x14 by variable _textY and store in r0x05
; ;Unrolled 8 X 8 multiplication
; ;FIXME: the function does not support result==WREG
BANKSEL _textY
; .line 153; oledWrite.c screenY = textY * 20 ;
MOVF _textY, W, B
MULLW 0x14
MOVFF PRODL, r0x05
MOVFF PRODH, r0x06
Perhaps the version of the compiler I'm using supports it and your version
doesn't?
I have:
sdcc -v
SDCC : pic16 2.9.0 #5416 (Oct 21 2009) (UNIX)
Gordon
------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
_______________________________________________
Sdcc-user mailing list
Sdcc-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sdcc-user
