On Wednesday 20 October 2004 12:17 am, Jan-Erik Soderholm wrote:
> Well, earlier you wrote :
> > if (foo[bar++] > 3) <-might be nice on a chip with pre/post
> > inc pointers such as 6809 but probably comes out
> > horrible on PIC.
>
> I was just asking why that whould "come out horrible
> on a PIC".

I was thinking of a certain x86 processor doing far pointers using 16bit 
commands when I said that, and therefore expected to see a fair bit of 
register swaps using dx,ax,ds and more.
I was also thinking about some rather large embedded programs which produced 
about 8k of code and did a lot of Freg swapping....
example routineX uses F10,F11,F12 as temp, but routineY also uses F10,F11, so 
a couple routines call both routineX and routineY, therefore you got a fair 
bit of swapping in/out of F10,F11 due to the conflicting routines (problem 
due to not being able to hold values on a stack).

(just guessing steps required since got no compiler to verify)
	movfw	bar
	incf		bar,f
	addwf	foo,w
	movwf	IND0
	etc...
which is reasonable...

Basically, call that statement I made earlier, a mistake on my part.
I shouldn't apply large ideas to a PIC since it's going to deal with tiny 
tables & re-use some registers within a limited prog + mem space anyways.

BTW, since you're hanging on that statement ready to pound it into the 
ground, I'm guessing you looked at it and got a ready reply...
....so what did your compiler produce?
_______________________________________________
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist