Shaky screens

Madskillz · Post by **Madskillz** » Sun 06 Feb, 2005 3:15 am

Well my 84+ SE never had this problem...except with phantom star, and I believe the old zDoom version...I talked with gH and he fixed it in the latest version: http://joepnet.com/hosted/maxcoderz/php ... .php?t=124

Yes, it was those darn enhanced fastcopy routines...from what I have seen and tested...if you use the plain old fastcopy or Duck's gsCopybuffer you should have no problems with the jumpy screen. But all those games that you listed, except for Phantom Star work just peachy on my calc. Hmm?

Jim e · Post by **Jim e** » Sun 06 Feb, 2005 3:41 am

I'm starting to have second thoughts about buying an 84+se, my 83+se works great and no shaky screens. I don't know, I have 150 bucks in my wallet and don't know if I want to waste 130 of them on a calc with a broken lcd driver.

Maybe I should wait till college

You know, I don't think trying to optimize all the copy routines is really offering that much any more if they keep buggy out. Maybe we should all convert to copy routines that wait indefinately.

Code: Select all

Lcd:	
	di
	ld c,$10
	ld a,$80
startwait:
	in (c)
	jp m,startwait
	out ($10),a
	ld hl,gbuf
	ld de,12
	ld a,$20
col:
	in (c)
	jp m,col
	out ($10),a
	ex af,af'
	ld b,64
row:
	ld a,(hl)
rowwait:
	in (c)
	jp m,rowwait
	out ($11),a
	add hl,de
	djnz row
	ex af,af'
	dec h
	dec h
	dec h
	inc hl
	inc a
	cp $2c
	jp nz,col
	ret

Sorry...there was a lack of code in this thread and I thought it needed some

CalcKing · Post by **CalcKing** » Sun 06 Feb, 2005 5:42 am

Jim e wrote:Sorry...there was a lack of code in this thread and I thought it needed some

Yes! Every thread should have some Z80 in it, so that people may learn by example!

As for the problems with lcd timings, it really would be nice if MirageOS 1.3 were released. But I haven't seen any sign of Dan in a while.

Spencer · Post by **Spencer** » Sun 06 Feb, 2005 6:14 am

TI's cheap decision in the past sure is screwing them over now. The routine above is indeed universal, but so is TI's grbufcopy (both unnecessarily slow, but given what we know about timings, may be neessary).

I just suggest putting 80+ clocks between writes. That should encompass all three 83+ family calcs.

Post by **tr1p1ea** » Sun 06 Feb, 2005 6:23 am

Yeah, well i wouldnt say that its screwing THEM over ... more like its screwing US over

.

I say 80+ is ok ... but would that work for the SE in 15MHz mode?

Jim e · Post by **Jim e** » Sun 06 Feb, 2005 7:45 am

15mhz...no, more like 160 tstates.
But mine is alot faster than ti's, so please don't compare.

they check bit 1 of port 2, that's a waste of clocks!!!
Compare:

Code: Select all

 in a,(2)
 bit 1,a                 ;TI's
 jr z,whereever

 in (c)                  ;mine
 jp m,somplace

I'm actually proud of that little bit of code.

CoBB · Post by **CoBB** » Sun 06 Feb, 2005 8:24 am

But you are checking bit 7 of the incoming data instead of bit 1. Does it actually work with the LCD driver?

Jim e · Post by **Jim e** » Sun 06 Feb, 2005 8:36 am

No...its checking the sign bit, which happens to be bit 7

but most may need to add the instruction in (c), aka in f,(c), to their tasm table becuase it's undocumented.