LCD SPEED misleading...

[Jim e]

Ok now I'm angry. I've been Using Michael Vincent's port reference to help me out with the grey scale thing. What was one of my favorite ports was the lcd speed. Increase the value in this port affects the needed delay for writing to the lcd, this is still true. However it doesn't increase the lcd controller speed, as Michael implied, it decreases cpu speed when using ports $10 & $11. Simplly put If I sent $FF to the port and OTIR the whole screen it wouldn't take 20,000 clocks, no it would take 87,000 clocks.
I gathered a little more info on port $29. Basicly, whatever is in port $29 divide by 4 and add to the clocks needed to 'out' , 'outi' , or 'otir'. That will be more accurate.
I tested this by out ($11),a 64 times and looped 40,000 times so it would normally take 5 seconds. I ran it with $ff in the port $29, it took 32 seconds. That pretty significant. I then replace OUT ($11),a with Add hl,hl which should have the same clocks but when I ran it, it took about 5 seconds.
So basicly, no matter the routine i use, it still takes about 54,000 clocks.
Sorry, I needed to vent.

[tr1p1ea]

I have also messed around a little with be-ridding the need for a delay with the LCD (using OTIR in a fastcopy routine) but i dont have an SE so i cant really tell if it better or not. Made a few graysacle examples but i dont think any were that great.

Also i have spoken to Michael_V about it and he claims that after testing it himself the cpu speed did not decrease?

[Jim e]

The way I tested was pretty straight forward.

Code: Select all

start:
	di
	ld a,$80
	out ($10),a
	ld de,40000   ;based on clocks it takes 5 seconds
loop:
	ld a,$20
	out ($10),a
	xor a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	out ($11),a
	dec de
	ld a,d
	or e
	jp nz,loop
	ret

using calcsys set port $29 to whatever the standard is, michael says its $17, mine resets to $14.
then just asm it at the homescreen, by the predefined clocks it should take 5 seconds, VTI confirms that. when testing on my calc it takes 7 seconds.
Now set the port to $F7(no delays required), it took me 32 seconds. No change in code just the port setting, I might be off by a second or two but not 32.
Unless my calc and my calc alone is messed up, I can't see any flaws in my logic.
maybe it doesn't affect the cpu speed but it does at least add a delay when outing. But if it is my calc I think it's about time for an 84.

[tr1p1ea]

Hhhmm... Thats actually an interesting discovery. Ill have a chat to Michael_V.