Problematic Demos.

[amilo3438]

@mc6809e

Well, I never said I would give up ... what I meant to say is that I also have need to see a day light, agree?

What I have been doing is finishing with testing of "not so popular" demos.
Hope that other "least popular" is not needed to be tested as it looks to be not so demanding .... and there are thousand of such demos in pouet exist.
Anyone would never care about those anyway.

Also, would be great in future if I would be not alone in this mission as only one who cares, agree?!

Hope this thread is also open for anyone who wish to help and not just only for me, agree?!

[amilo3438]

Originally Posted by mc6809e

Do you have a working A500? If you don't then perhaps you can get other users to verify the suspected error on a real machine. Then the demo with documented error can later be presented for correction.

Yes I have an A500 '91/'92 version (bought in Feb '92), but... I am currently running a laptop that has no any floppy drive, only USB ports (my old desktop is still broken).
And till today I didnt find a way how to do a Amiga demo/floppy transfer using USB ?! (Without buying an USB floppy drive.)

So would be a really great if someone else would help to done those testings.

p.s.
My old way was using 720kb PC drive that an Amiga can read ... it worked but as remember it was also very slow way.

[mc6809e]

Originally Posted by amilo3438

.

Also, would be great in future if I would be not alone in this mission as only one who cares, agree?!

Hidden in your statement is the suggestion that people are not doing what you think they should be doing. You sound like my mother-in-law. She says little things that create an implied burden on others, then is surprised that people are annoyed with her.

Please don't do that. We're here for fun. We're not here to perform any kind of duty or to satisfy some obligation. If you get people to join you in your mission, then great. If you try to pressure people to act then you'll just get resistance.

It's like anything else. To volunteer is rewarding. Work is miserable. Don't turn the maintenance of WinUAE into work.

Keep the experience fun for everyone.

[Toni Wilen]

Hotbleeps: This is strange.. 2.6.1 at least works correctly, looks exact same as on my A500.

Pouet screenshot is wrong, video is wrong, even the included screenshot with archive is wrong! (What happened here?)

[amilo3438]

Originally Posted by mc6809e

Hidden in your statement is the suggestion that people are not doing what you think they should be doing. You sound like my mother-in-law. She says little things that create an implied burden on others, then is surprised that people are annoyed with her.

I agree, and you're probably right, but those words comes out as I've got impression that till a few ours ago I am being only one active here ... and you know, the World is big, so big ... but here is only me, how strange ... but it looks not anymore as I've got a new thread-friends here.

p.s. Please say Hello to your mother-in-law ... it was very funny reading this.

[Toni Wilen]

I think some posts are not really on topic anymore..

Prurient by Rebels: off-by-one bug in recent sprite update (260b15). For some reason left (which is missing now) and right white line are not exact same, right side line misses top and bottom pixel. A500 confirmed.

[StingRay]

Originally Posted by Toni Wilen

Hotbleeps: This is strange.. 2.6.1 at least works correctly, looks exact same as on my A500.

Pouet screenshot is wrong, video is wrong, even the included screenshot with archive is wrong! (What happened here?)

Just a guess: intro has been coded and tested on UAE only and due to a bug in the emulator it looked different in an older version? I'll ask Blueberry next time I meet him.

[amilo3438]

Originally Posted by Toni Wilen

Freedom Force Megademo: Not emulation bug. It is very common overscan problem. (Usually real hardware visible only if using monitor that allows size adjustments)

Just for info...

Found some other interesting examples like on NorthStar Megademo3:

On the 1st picture is visible 2 different shifts of green lines.
What could cause that these shifts are different?!

On attached 2nd picture there is no any shifting visible on red lines.
Both cases are same demo running on WinUAE window resolution 752x576.

Could that be something related to cooper instead overscan?! just wondering

p.s. Used tested files were named "NST_FLT-Megademo3A.dms" and "NST_FLT-Megademo3B.dms".

[Toni Wilen]

Originally Posted by StingRay

Just a guess: intro has been coded and tested on UAE only and due to a bug in the emulator it looked different in an older version? I'll ask Blueberry next time I meet him.

Possibly but imho it looks more interesting now

Originally Posted by amilo3438

Just for info...

Found some other interesting examples like on NorthStar Megademo3:

On the 1st picture is visible 2 different shifts of green lines.
What could cause that these shifts are different?!

On attached 2nd picture there is no any shifting visible on red lines.
Both cases are same demo running on WinUAE window resolution 752x576.

Could that be something related to cooper instead overscan?! just wondering

Yes, bad copper wait values. Wait for hpos=1, write to background color, color change is visible in right border. Correct way is to wait for larger hpos value before changing colors.

Hblank starts at hpos=7 or so, not at hpos=0.

[Toni Wilen]

Originally Posted by amilo3438

COOL FRIDGE by UPFRONT -> http://www.pouet.net/prod.php?which=1467

Running on Quickstart A500 OCS/ECS glitches on "SMOOTH COPPER" part an UPFRONT logo only when is copper plasma behind (see attach pict. 1 & 2).

We have a winner here! This has to be the most ugliest and weirdest way to syncronize CPU to display..

It WRITES to VPOSW to force CPU vs chipset sync! It writes to chipset registers in a loop, then increases vpos counter and writes it to VPOSW and then starts processing next line.

WTF? Why? This is too crazy..

I guess timing is close enough to real hsync and TV allows small jitter in hsyncs without causing display glitches.

I don't think I want to bother with this. It would require huge changes everywhere..

Code snippet:

00065566 4e71                     NOP.L
00065568 33c0 00df f02c           MOVE.W D0,$00dff02c
0006556E 4e71                     NOP.L
00065570 4e71                     NOP.L
00065572 4e71                     NOP.L
00065574 4e71                     NOP.L
00065576 33c1 00df f02c           MOVE.W D1,$00dff02c
0006557C 0641 0100                ADD.W #$0100,D1
00065580 0640 0100                ADD.W #$0100,D0
00065584 0642 0100                ADD.W #$0100,D2
00065588 0643 0100                ADD.W #$0100,D3
0006558C 4e71                     NOP.L
0006558E 4e71                     NOP.L
00065590 4e71                     NOP.L
00065592 4e71                     NOP.L
00065594 4e71                     NOP.L
00065596 4283                     CLR.L D3
00065598 4e71                     NOP.L
0006559A 4e71                     NOP.L
0006559C 4e71                     NOP.L
0006559E 4e71                     NOP.L
000655A0 4e71                     NOP.L
000655A2 4e71                     NOP.L
000655A4 33c0 00df f180           MOVE.W D0,$00dff180
000655AA 4e71                     NOP.L
000655AC 4e71                     NOP.L
000655AE 4e71                     NOP.L
000655B0 4e71                     NOP.L
000655B2 33c1 00df f180           MOVE.W D1,$00dff180
000655B8 0641 0100                ADD.W #$0100,D1
000655BC 0640 0100                ADD.W #$0100,D0
000655C0 0642 0100                ADD.W #$0100,D2
000655C4 0643 0100                ADD.W #$0100,D3

[StingRay]

"Nice" code indeed. I'm almost sure the coder used to code C64 demos before as the code looks more or less like 6510 code converted to 68000.

[amilo3438]

Originally Posted by Toni Wilen

We have a winner here! This has to be the most ugliest and weirdest way to syncronize CPU to display..

It WRITES to VPOSW to force CPU vs chipset sync! It writes to chipset registers in a loop, then increases vpos counter and writes it to VPOSW and then starts processing next line.

WTF? Why? This is too crazy..

I guess timing is close enough to real hsync and TV allows small jitter in hsyncs without causing display glitches.

I don't think I want to bother with this. It would require huge changes everywhere..
[/code]

Indeed very interesting case. Also bad that it would require such huge changes to fix.

p.s.
I have checked all demos again ... hope you did not overlook below by mistake testing on real Amiga:

The end by trilobit, Vactro 2 by Angels, ALUMINIUM by SYNTEX

as didn't find anywehere you mentioned these. I'll be here in case you'll need any more info about.

-----------------------------------------------------------------------------------------------------------------------------------------
EDIT: Maybe could be an useful info: COOL FRIDGE by UPFRONT demo ... on SMOOTH COPPER loading screen it says (see on attachments also):

LOADING 'SMOOTH COPPER' , NOTICE THAT
THE SPLITS ARE MOVING ON EVERY SECOND
PIXEL BEHIND A 4 PLANE LOGO....

[Toni Wilen]

Originally Posted by amilo3438

Vactro 2 by Angels

Flickers 100% exactly same on real A500..

[Toni Wilen]

Originally Posted by amilo3438

ALUMINIUM by SYNTEX -> http://www.pouet.net/prod.php?which=3027

Quickstart A500 ... after some time of running demo only black screen shows.

Which is correct behavior. Same happens using real A500. Including corrupt scroller at the end if some fast ram is added (to allow raytrace part to work).

Some more testing: It seems to work fully if KS 3.1 and some fast ram enabled..

[amilo3438]

Silly scroller by Phenomena -> http://www.pouet.net/prod.php?which=18018

A500 OCS/ECS ... strange small white line appears?!


p.s. For info ... have just finished testing Amiga OCS/ECS pouet collection (at least those demos that have been rated with double-digit thumbs up).
So theres now left only cracktro and dentro collection to test ... at least those better rated.

[amilo3438]

Originally Posted by Toni Wilen

Hotbleeps: This is strange.. 2.6.1 at least works correctly, looks exact same as on my A500.

On pouet nfo (or lha archive readme) is said that is for an Amiga OCS (only)?!
Have you been tested it on OCS or ECS Amiga ?

Originally Posted by StingRay

Just a guess: intro has been coded and tested on UAE only and due to a bug in the emulator it looked different in an older version? I'll ask Blueberry next time I meet him.

Could be, as on pouet is said release date june 2011 ... I have been tested it on WINUAE2410 (1st half of 2012 version?!) and was looked exactly same as on picture or video.

EDIT: But also a little strange is that it worked in WinUAE2410 (and earlier versions) A500 regardless OCS, ECS Agnus, Full ECS or AGA was used (cause nfo file said its for Amiga OCS only ?!).

EDIT1: But in WinUAE2410 if change from PAL to NTSC got different screen look similar to latest WinUAE versions ?!

[Toni Wilen]

Originally Posted by amilo3438

On pouet nfo (or lha archive readme) is said that is for an Amiga OCS (only)?!
Have you been tested it on OCS or ECS Amiga ?

ECS. I am not interested in swapping Agnus chip because I am 100% sure it won't have any effect. It only depends on copper speed.

EDIT1: But in WinUAE2410 if change from PAL to NTSC got different screen look similar to latest WinUAE versions ?!

It is normal. PAL and NTSC has different number of DMA cycles/scanline. (PAL is always same, NTSC alternates)

[Toni Wilen]

Prurient fixed (stupid off-by-one bug)

Silly scroller finally fixed.

This was really difficult to debug because it was not actual chipset emulation bug but internal emulation bug (internal buffer was not always cleared correctly, this is another speed optimization, only clear just enough, never too much) when resolution or number of bitplanes change mid-scanline. For some reason this didn't trigger with Disposable Hero or others that really abuse BPLCON0 mid-scanline.

This is one of those demos that have (most likely accidental) BPLCON0 write mid-scanline. I am quite sure coder was confused, he probably thought it was some HRM or chipset bug, then he decided to adjust bitplane pointers and/or DDFSTOP/STRT until display was fixed.

RAF Megademo has similar problem. BPLCON0 is written too late, few cycles after DDFSTRT has matched. It causes first 2 bitplane fetches to be ignored. (2 because it is hires and OCS/ECS "display unit" is always 8 cycles, emulation bug caused second half of "display unit" to not be skipped in hires mode if first half didn't start) Unfortunately there is also some Denise internal feature that adds another shift (about 2 pixels) which is not yet fixed. (I don't have any ideas yet)

Coder actually had to add extra code to work around the shift in first line! (Instead of fixing the copper list)

[amilo3438]

Reminder on comments and current status

+ Blerkenwiegel by Scoopex -> (Broke in 2.6.1b1. It was accidental side-effect caused by another tiny change..) -> Fixed!?

+ Funky Room by Y-Crew -> (99.9% guaranteed not-OCS compatible problem. Sets ECS only DDFSTOP bit.) -> 99.9% not-OCS problem!

+ The end by trilobit -> (Added CIA tod bug option to Advanced chipset.) -> Fixed!

- Hotbleeps by Loonies -> (This is strange.. Pouet screenshot is wrong, video is wrong, even the included screenshot with archive is wrong! What happened here?)

+ Prurient by Rebels -> (Off-by-one bug in recent sprite update (260b15)) -> Fixed!

+ Vactro 2 by Angels -> (Flickers 100% exactly same on real A500..) -> Same as on a real A500!

+ ALUMINIUM by SYNTEX -> (It seems to work fully if KS 3.1 and some fast ram enabled..) -> Work with KS 3.1!

+ Antediluvian Sloppy Spectacle by essence -> (Usual "we don't disable sprite dma during vblank" bug.) -> Fixed!

- COOL FRIDGE by UPFRONT -> (We have a winner here! This has to be the most ugliest and weirdest way to syncronize CPU to display..)

+ Megademo by Royal Amiga Force -> (Denise internal feature that adds another shift (about 2 pixels) which is not yet fixed.) -> Fixed! (2013-11-02)

+ Megademo by Freedom Force -> (Not emulation bug. It is very common overscan problem.) -> Not emulation bug! http://eab.abime.net/911751-post80.html

+ Silly scroller by Phenomena -> (..it was not actual chipset emulation bug but internal emulation bug..) -> Fixed!

+ Made In Croatia by Binary -> (Same as Prurient.) -> Already Fixed!

+ Subtle shades by Nuance -> (It is ECS only, OCS Agnus = corruption. old post) -> ECS only!

+ 5o2 by Fairlight -> (Not interested, it is already confirmed broken in pouet link.) -> Confirmed broken!?

+ Trashcan by Wizzcat -> (tested versions are damaged) -> Working version! http://eab.abime.net/911671-post70.html (see link on EDIT2)

+ Tales Of A Dream by Dreamdealers-> (Usual bad copper wait. There are hundreds of those.) -> Not emulation bug! http://eab.abime.net/911751-post80.html

+ Shave Off Your Moustache by The os -> (tested DMS version is damaged) -> Working version! http://eab.abime.net/911684-post73.html (see link on EDIT)

- Pax Americana by Complex -> (It writes to blitter registers while blitter is active ... impossible to fix without having blitter's original logic diagrams/schematics.)

- Spiritual connection by Amaze -> (Another writing to blitter registers while blitter is already doing stuff.)

+ Starflight by Phenomena -> (Now fully emulated, scroller will jump correctly in ECS mode.) -> Fixed!

- Critical Mass by Parallax -> (can happen on real hardware too but appears to be more rare, probably some internal CIA timing not yet fully emulated)

- Application-Trackmo -> (BLITTER: register modification during linedraw!)

+ Megademo by Trilogy -> (A500 confirmed. No idea why sprite DMA is enabled or pointers point to garbage) -> Same as on a real A500!

+ Copper Slave by Ram Jam -> (Modifying bitplane modulo one cycle before modulo is added will use old value.) -> Fixed!

+ Corsair's Latest by the company -> (exact same fill leak happens on my real A500, including quite jittery speed. Broken executable?) -> Same as on a real A500!

+ Rapiad by Byterapers -> (Same corruption seen on my real A500.) -> Same as on a real A500!

+ Freedom by Nova -> (Only did real A500 ECS test and it has same glitches. I assume it works on OCS A500.) -> Same as on a real A500!

+ Little Mega Demo by Mad Monks -> (A500 confirmed. It starts bitplane DMA 1 scanline before setting bitplane pointers = shows random memory contents, possibly nothing visible in some configurations.) -> Same as on a real A500!

+ New Stuff by The Special Brothers -> (ECS A500 and A1000 confirmed. Need another version to confirm possible corruption.) -> Same as on a real A500!

+ Megademo by Intuition -> (Not confirmed but most likely emulation problem. Not exactly same problem as in RAF demo) -> Fixed! (2013-11-02)

+ No Reality by Devils -> (Fixed. CE mode CMPA instruction was 2 cycles too slow.) -> Fixed!

+ Nanotech Mysteries 2 by Limited Edition -> (A500 confirmed. Quite weird because overscan isn't that large in this demo.) -> Same as on a real A500!

+ Tales of Mystery and Imaginations by Syntex -> (Make sure DF1: is unticked. Note that CLI prompt _should_ appear, it will still continue loading.) -> Same as on a real A500!

+ ECA Demo 2 (Demo) by Electronic Cracking Association -> (A500 confirmed. One sprite pointer points to garbage for some reason.) -> Same as on a real A500!

+ Oz Con Placings by Alchemy -> (Not bothered check because they are 100% surely standard interlace artifacts when shown on progressive screen.)

+ Gerp 2013 Invitation (Oldskool Demo) by Nature & Traktor -> (Yes, it flickers on real A500 too. I think it will work fine with (real) 68020+ due to bigger prefetch buffer.) -> Same as on a real A500!

+ Anti Ivory Demo by Iq Minus -> (Yes and it really does look very glitchy, including that blue garbage at the bottom part of screen.) -> Same as on a real A500!

+ AHej Hopp (aka Goofys 3rd demo) by The Silents & Northstar -> (It is part of design. It simulates crash.) -> Same as on a real A500!

+ Demo 1 by The Link -> (Some weird sprite garbage. Does not happen on ECS A500 (or A1000)) -> Fixed!

+ Delirium State of Mind by void -> (This is not A500 demo (or it is quite bad one.)) -> Same as on a real A500!

+ Scrollnophobia by Proton Ltd -> (Usual overscan bugs.) -> Same as on a real A500!

+ Awake On The Wildside by Network -> (Usual overscan bugs.) -> Same as on a real A500!

+ ABurger Hill by D-Tect & Coma -> (It is a joke!) -> Same as on a real A500!

+ Bergen Party Demo by Cartel -> (Different file lengths, different contents. But anyway, A500 confirmed.) -> Same as on a real A500!

+ Animotion by Phenomena -> (Gap in background color bar seems to be ham mode-related emulation bug.) -> Fixed!?

- Equipose by Complex -> (It is a bug in demo that accidentally works on real hardware.)

+ Faketro (Intro) by Scoopex -> (A500 confirmed. (Why? I have no idea)) -> Same as on a real A500!

+ VektoriJa Viuhahdus (Intro) by Complex & Scoopex -> (Problem is not a problem. It is supposed to leak like that.) -> Same as on a real A500!

+ Geil Intro (Intro) by Alpha Flight -> (working fine in beta 7 ... it starts be broken in current beta 8) -> Fixed!

+ Compass 1 (Diskmagazine) by Alcatraz -> (Same problem. Scrolling has nothing to do with it.) -> Fixed! (2013-11-02)

- Madness 2 - The Search For Sanity by Anarchy -> (This kind of "effect" is very common when "something" takes too long or is too quick and program requires very good timing and is "racing the beam".)

+ Master Of Magic Demo by Anarchy -> (A500 confirmed (OCS), ECS = very corrupted display because it triggers chipset bug, this does not happen in emulation but I added log warning.) -> Same as on a real A500!

+ Hammer's Mini Intro by Anarchy -> (work fine on A500 OCS 0.5mb chip ram)

+ Deja Vu by Anarchy -> (Fixed, some 68020 stuff leaked to 68000 side..) -> Fixed!

+ Plasma Vectors by The Silents -> (A500 confirmed.) -> Same as on a real A500!

+ Check Mate Intro by The Silents -> (A500 confirmed.) -> Same as on a real A500!

+ Misery Dentro 2 by TRSI & Masque -> (A500 confirmed.) -> Same as on a real A500!

+ Hi Freaks (Intro) by Supplex & The Silents -> (Corruption happens because CPU modifies bitplane pointers in copperlist after copper has fetched high word but before copper fetches low word of bitplane.) -> Updated 2013-10-27 ... (seems to work fine on beta11 2013-11-02)

+ Necromantic Clouds by TRSI & Cachet -> (A500 confirmed. I assume this demo was made for faster Amigas.) -> Same as on a real A500!

+ Forensic Clinicism (40k Intro) by TRSI -> (It looks like it is demo bug (uninitialized variable or something).) ...!?!

+ Primus (Demo) by Brainstorm -> (invalid refresh rate -> result is undefined -> emulation is fine) ...!?!

+ Protracker 3.00 beta - intro (Intro) by Cryptoburners -> (A500 confirmed, exact same glitches. Probably made for faster machines.) -> Same as on a real A500!

+ Seven Months Old (Demo) by The Crusaders -> Fixed!

+ Switchblade's Balls (Demo) by The Crusaders -> (A500 confirmed. Shows uninitialized memory.) -> Same as on a real A500!

+ Blue Lagoon (Demo) by Wizzcat & Belga-live -> (A500 confirmed and this is "most common" side-effect on buggy demos.) -> Same as on a real A500!

+ We shave Ass (Demo) by Razor 1911 -> (Broken change in test version, also broke other similar scrollers.) -> Fixed!

+ Mr. Peq's Amiga First (Demo) by Dual Crew -> (Some tweaks done but not guaranteed to be correct, this is quite rare situation.) -> Fixed!

+ Now Thats What I Call Crap (Demo) by Lsd -> (Just another demo with "scrolling text" issue!) -> Fixed! (2013-11-02)

+ Pleasure Trip (Demo) by Lsd -> (80 character mode required.)

- Funky Big-Mac (Intro) by Lsd -> (Need to examine this more..)

+ Delirka (Demo) by Appendix -> (It is AGA demo.)

+ Intel Inside (Intro) by Appendix -> (That error means not enough memory.)

+ Boot Intro by Tristar -> Same as on a real A500!

+ BBS Intro by Spreadpoint & Amiga Industries -> (It has been crunched with Imploder's library mode and will not run without explode.library!)

+ First Release in 1987 (Intro) by Amiga Industries & Kingsoft -> (80 character mode required.)

+ Digital Innovation (Trackmo) by Anarchy -> (Experimental test code remaining in b9.) -> Fixed!

+ Delirium (Demo) by Complex -> Same as on a real A500!

- Shed Tears by Ozone -> (This can't be fixed (without having real interlaced display)).

+ Alpha and Omega (Demo) by Pure Metal Coders -> (MoreNewStuffy / PlasmaForce was non-modulo case, this one is modulo case.) -> Fixed!

+ Deformations by Deform -> (This demo has buggy trackloader.)

+ Jointro (Intro) by Analog -> Same as on a real A500!

+ Short and Fast Production (Intro) by Analog -> Same as on a real A500!

+ Meeting Demo (Demo) by Celtic -> (It seems it requires KS 1.2 or older.) -> Same as on a real A500!

+ JMcDisk 2 Intro (Intro) by Coma -> (This is weird one. Background is black with KS 1.1, 1.2+ and top 3/4 is blue, bottom part is black.)

+ Megademo by Cult -> (It seems it requires KS 1.2 or older.) -> Same as on a real A500!

+ Born (Demo) by Cult -> (Seems to be yet another KS 1.2 demo.) -> Same as on a real A500!

+ New and Better Routines (Intro) by Cult -> (adjusted but it probably still isn't exactly right, needs more testing, someday.) -> Fixed!

+ Boot Intro by Digitech -> (This is probably bad bootblock rip that does not "emulate" bootblock environment fully.)

+ Neon Skates Demo by End Of Century 1999 -> (It needs mathtrans.library.)

+ Simple A Demo by Gate -> (A1000 Agnus does not enable bitplanes if vertical DIW is zero. Only 1 and larger works.) -> A1000 Fixed!

+ Metallpower (Demo) by Powerdrive -> Same as on a real A500!

+ Balls & Motorboats (Demo) by Savage -> (68000/68010 cpu core merge had a bug, some 68000 special instructions lost their 68000 special behavior.) -> Fixed!

- Red It Intro by Savage -> (A500 ECS: logo is white A1000: logo is black. Reason unknown.)

+ Wonderboy's First (Intro) by Vortex 42 -> Same as on a real A500!

- Anarchy Party Demo by Symbiosis -> (No ugly dots on A500.)

+ Ting Man Aldrig Maa Goere (Demo) by Rednex -> (White flash is there, but position is random on real A500.)

+ Amiga Expo 92 (Slideshow) by Trygon -> Same as on a real A500!

+ New Demo by British Swapping Syndicate -> (I can't see any obvious difference on my desktop (overclocked 4770k), probably need to test it on slower PC...)

+ Marchewki (Trackmo) by Alchemy -> (A500 confirmed. (Corrupted data?)) -> Same as on a real A500!

+ Lethal Formula Final (Trackmo) by Cult -> ( A500 confirmed. (Single line of hidden surface being visible)) -> Same as on a real A500!

+ Alternative Reality (Demo) by Suspect -> (A500 confirmed. (Corrupted data?)) -> Same as on a real A500!

- Last A500 (Demo) by Oblivion -> (most likely caused by some CPU emulation timing problem.)

+ Feel Blue (Intro) by Oblivion -> (Another scrolling text issue.) -> Fixed! (2013-11-02)

+ Megademo by Sargon & Red Sector Inc. & Thrust & Phalanx & Shadow Light -> (This is OCS only. ECS = single mismatched row. Emulation slightly tweaked to match this behavior.) -> Fixed!

+ Cheap demo (Demo) by Phoenix -> (A500 confirmed, shows same garbage.) -> Same as on a real A500!

[amilo3438]

Made In Croatia by Binary -> http://janeway.exotica.org.uk/release.php?id=9158

This one looks very similar to Prurient (off-by-one bug) problem, thus might happen that is already fixed?!
But anyway, that will need to be checked to be 100% sure.