Problematic Demos.

[ross]

I can confirm, there can be small potential problems with the intro initialization code and I understand why in rare cases it can crash on startup.

However even if it has nothing to do with the crash I'd be curious to know why the coder used that weird manual interlace setting via VPOSW

Originally Posted by rutra80

No I start it via mounting the archive as floppy. I might record video of how it is - are you interested in A500, A1200 issue, or both?

Well, if it's not a problem, post the video(s).
But are there different issues? Or is it the same for A500 and A1200?

However I really don't think they are problems related to the emulation, but to some display settings (or due to the 'inversion' of the LOF in the intro).

[Toni Wilen]

Seems like usual bug, KS vblank interrupt is running and if KS 3.1+ (and no system level interlace screen active): LOF bit is set in KS routine.

Not sure if LOF "detection" position was confirmed (or it was display specific?)

(2023 and there are still wrong right border copper waits..)

[rutra80]

Here are some explanations:
https://www.pouet.net/prod.php?which=94588

My A500 config seems to have no issues from what you say, I'll post video of my A1200 config with unexpected scanlines and supposedly LOF inversion problem.

EDIT: Toni - if quickconfig for A1200 uses KS3.0 then it might be it - my manual A1200 config uses KS3.1 IIRC (though would inversion of the LOF generate scanlines?), I'll post more in the evening...

[ross]

Originally Posted by Toni Wilen

Seems like usual bug, KS vblank interrupt is running and if KS 3.1+ (and no system level interlace screen active): LOF bit is set in KS routine.

Yep, this, plus more.

Originally Posted by rutra80

if quickconfig for A1200 uses KS3.0 then it might be it - my manual A1200 config uses KS3.1

Yes, this.

But maybe a full fix for the intro is better.
It's not complicated, I'll do it tonight.

[ross]

Into the Zone! Capsule-SyntroN fix.

Patches:
- AGA and ECS proper recognition
- KS31 bad LOF fix
- No more crash on KS13 and NTSC
- Sine scroll centered
- border as it should be
- some fixex on copper list
- a bit less glitchy (not much, there is a memory leak somewhere)

It should work on any configuration...but i haven't tested it much.
Please test it.

[rutra80]

Ok everything confirmed - there's LOF inversion problem on KS3.1
I think there's no need for video as it's enough to switch A1200 quickconfig to KS3.1 and Fullscreen with proper scaling (integer) and you shall see that scanlines of unknown to me origin.

[ross]

Originally Posted by rutra80

Ok everything confirmed - there's LOF inversion problem on KS3.1
I think there's no need for video as it's enough to switch A1200 quickconfig to KS3.1 and Fullscreen with proper scaling (integer) and you shall see that scanlines of unknown to me origin.

Does the fix work correctly?

I forgot to mention that I also eliminated the wacky VPOSW interlacing for a normal BPLCON0 LACE bit

[rutra80]

KS: 3.1
Line mode: Double
Interlaced line mode: Double frames
Scaling: Integer

All seems fine except that isometrics seem borked:

[ross]

Originally Posted by rutra80

All seems fine except that isometrics seem borked:

It doesn't 'seems', it is.
It depends on where it loads the hunks, or the use of fast-ram, and there are (little) differences.
It's the memory leak I was talking about, it's lucky that it doesn't crash or glitch very badly.

But I won't fix that, in case the coder will take care of it
In any case he clearly specified that it is not a finished product, so that's okay.

[rutra80]

Yeah too many bugs making me wonder if it's intro or WinUAE, everything is explained now. Hope at least you had some fun with it

[Toni Wilen]

Originally Posted by ross

Into the Zone! Capsule-SyntroN fix.

Patches:
- AGA and ECS proper recognition
- KS31 bad LOF fix
- No more crash on KS13 and NTSC
- Sine scroll centered
- border as it should be
- some fixex on copper list
- a bit less glitchy (not much, there is a memory leak somewhere)

It should work on any configuration...but i haven't tested it much.
Please test it.

Real unexpanded KS 3.1 A1200 results:

Original: OSSC won't sync, OSSC display flickers between two different refresh rates (impossible to see values because display updates continuously)

Fix: works, middle isometric cube structure part looks ugly (swapped fields?). Otherwise it looks ok. Bottom scroller first few topmost lines jumping horizontally slightly, the rest of scroller is smooth. Also some random pixels appears and disappears mid screen, these don't appear in UAE.

Looks almost identical to A1200 WinUAE in fields+ mode.

[amilo3438]

Go 64! v2.0 - import (Crack Intro) by World Of Wonders => http://janeway.exotica.org.uk/release.php?id=23517

Check the intro "WOW-CrGo64V2.adf" file from the above link...
There is a boot block and it seems WinUAE has some problem with F7 PAL <=> NTSC option !? (it freezes)



264 Line Object (Demo) by Northstar => http://janeway.exotica.org.uk/release.php?id=14124

Some distortion in the music after a while! (need to listen a little longer)



2nd Demo (Demo) by Mafia => http://janeway.exotica.org.uk/release.php?id=89678

Some top line glitches on the screen! (QS 1.2 OCS 512) EDIT: No glitches if disk is inserted after WB floppy screen!
Ps. No glitches on Denise! (same config)

[ross]

Originally Posted by amilo3438

Check the intro "WOW-CrGo64V2.adf" file from the above link...
There is a boot block and it seems WinUAE has some problem with F7 PAL <=> NTSC option !? (it freezes)

This is due to some dumb code that expects a custom write-only register to return 0 as the value (write-only registers must not be read, content is undefined!).
WinUAE return $FFFF and the very same value is written in BEAMCON0 -> NOSYNC video state.

[amilo3438]

Originally Posted by ross

This is due to some dumb code that expects a custom write-only register to return 0 as the value (write-only registers must not be read, content is undefined!).
WinUAE return $FFFF and the very same value is written in BEAMCON0 -> NOSYNC video state.

I wonder how it behaves on a real machine? (maybe the solution is to make it to return 0 as the value)

[paraj]

Originally Posted by amilo3438

I wonder how it behaves on a real machine? (maybe the solution is to make it to return 0 as the value)

On A1200 it seems like just reading from $dff1dc modifies the register! On plain A1200

move.w $dff1dc,d0

sets NTSC mode! Likely "bus noise" is written, since move.w #$20,$dff1dc before and it's "fine". With 060 enabled monitor looses sync from plain write (again probably due to bus noise thing), if I do move.w #$20,$dff180 / move.w $dff1dc,d0 it's also fine. Move 0 to color0 and it changes to NTSC mode.

[ross]

Originally Posted by paraj

On A1200 it seems like just reading from $dff1dc modifies the register! On plain A1200

move.w $dff1dc,d0

sets NTSC mode! Likely "bus noise" is written, since move.w #$20,$dff1dc before and it's "fine". With 060 enabled monitor looses sync from plain write (again probably due to bus noise thing), if I do move.w #$20,$dff180 / move.w $dff1dc,d0 it's also fine. Move 0 to color0 and it changes to NTSC mode.

The ADF does a single bset #5,BEAMCON0, used as a toggle..
So nice

[Toni Wilen]

Reading write-only register always causes write. What CPU reads and what gets written to register (they are not same) depends on many things, for example what DMA or CPU did in previous cycle, hardware model, chipset model, CPU type. There is no simple generic solution.

EDIT: A500 68000 should have been mostly correct already, need to check this later..

[amilo3438]

New 3D Intro (Demo) by Phalanx ... http://janeway.exotica.org.uk/release.php?id=2117

JFI... Another bitplane/refresh cycle conflict side-effect! (bad only in ECS)


Demo Disk 51 Intro (Intro) by Ipec Elite ... http://janeway.exotica.org.uk/release.php?id=21764

Not sure what happened here!? (Note: From pack disks works!)

[PiCiJi]

Jim Power "Two Live Crew" crack hangs. it modifies the blitter while it's running and disables "Skip B" and "Skip Y" same time.
When this happens and Stage X is active, the crack intro expects "probably" Stage A to be active in addition to Stage Y.
At least another "shifter bit" is necessary so that the blitter counts down faster.

[Zarnal]

http://ftp.amigascne.org/pub/amiga/G...C-CrTrJimPower