Problematic Demos.

[vagrant]

Delirium by Complex
http://janeway.exotica.org.uk/release.php?id=8705

Tested with A500 quickstart.
There is flickering white corrupt gfx during the Star Trek vector part (see pic).
Disabling cycle exact fixes these scenes, however this needs to be done on the fly as the demo will crash in other parts with CE disabled.

[Toni Wilen]

Originally Posted by ED-209

Delirium by Complex
http://janeway.exotica.org.uk/release.php?id=8705

Tested with A500 quickstart.
There is flickering white corrupt gfx during the Star Trek vector part (see pic).
Disabling cycle exact fixes these scenes, however this needs to be done on the fly as the demo will crash in other parts with CE disabled.

I get nearly identical glitches on my real A500. It looks like it has timing issue in blitter interrupt handling.

EDIT: I am quite sure I saw glitch-free versions ages ago too..

[Toni Wilen]

Delirium (1991)(Complex)[a][A92#2] in tosec works slightly better (non-a has same problem), less glitches but some still remain.

My guess for the bug is this code sequence:

; enable blitter interrupt
00015256 3b7c 8040 009a           MOVE.W #$8040,(A5, $009a) == $00dff09a
; clear blitter interrupt request
0001525C 3b7c 0040 009c           MOVE.W #$0040,(A5, $009c) == $00dff09c
; set blitter interrupt request
00015262 3b7c 8040 009c           MOVE.W #$8040,(A5, $009c) == $00dff09c

I think this was supposed to "kickstart" blitter interrupt chain but it does it twice if blitter if blitter interrupt request is already active (and it is). Clear should be before enable.

[amilo3438]

Originally Posted by Toni Wilen

Delirium (1991)(Complex)[a][A92#2] in tosec works slightly better (non-a has same problem), less glitches but some still remain.

Interesting...

COMPLEX-Delirium.DMS (kestra bitworld) CRC=1AC2E2E3 (on download it says CRC: 798D4619 ... hmm, seems .adf was replaced with .dms version, but why !?)

Complex-Delirium.adf (arabuusimiehet) CRC=798D4619 (FAQ says: "Everything you see here is hand checked and verified to be working")

Delirium (1991)(Complex)[A92#2].adf (tosec) CRC=798D4619

Delirium (1991)(Complex)[a][A92#2].adf (tosec) CRC=65BC50C1


So, according to real Amiga testing CRC=65BC50C1 from tosec works the best!?


EDIT:

OK, after changing .DMS version to .ADF:

dms2adf.exe COMPLEX-Delirium.DMS COMPLEX-Delirium.ADF (new ADF CRC=798D4619)

...so, CRC=1AC2E2E3 .DMS version is exactly same as CRC=798D4619 .ADF version!!!


Also noticed one difference between two tosec versions:

the CRC=65BC50C1 version contains also one additional "vector part" with 3D COMPLEX logo rotation before the ship entrance into the base!


EDIT2: @ED-209

If you want to see the CRC=798D4619 demo running without glitches, use an old version of emulator, like f.e. WinUAE 2.0.1 (2009.12.23) !!! (just tested)

p.s.
Now its understandable how it happened that it sneaked through the "arabuusimiehet" demo collection even though was "hand checked and verified to be working"!
Yeah, it was running fine on an earlier (read less accurate) version of the WinUAE emulator in that time! (so, has not been tested on a real machine)

It certainly could not happen in today's WinUAE emulator for much greater accuracy!
So it would not be a bad idea to double-check all the previously tested demos, just in case! (if there is a new volunteer ready for the task)
(otherwise a great demo collection, just a pity that it was not checked on a real machine)

[vagrant]

Interesting.
Delirium (1991)(Complex)[a][A92#2] seems to be some sort of final version, with extra effect.. I wonder which version was run at assembly.
Either way this is easy the best version, still some minor corruption but nothing that can't be fixed with a few [ Show youtube player ]
Thanks for investigating!

[Foebane]

I'm getting used to emulators that refuse to work properly with one or two demos, or games or whatever, as it seems tricky to try and fix it to work with them only to break others. I've got a couple of A500 demos that have major glitches in them, still, but I just make do.

Heck, in the last week, I've been asking around the C64 community as to why a particular demo I've got doesn't go beyond a certain point on WinVICE or other emulators, but I guess I'll have to leave it.

[Toni Wilen]

Always mention names! It is also not too rare that some demos are actually broken (by design even!) and won't work with real hardware either.

[Foebane]

Originally Posted by Toni Wilen

Always mention names! It is also not too rare that some demos are actually broken (by design even!) and won't work with real hardware either.

OK, the C64 demo in question is "Dutch Breeze" by Blackmail, and as I said it doesn't fully work with WinVICE. Nor in fact, did it fully work with CCS64. It only ran via some newer emulator called Hoxs64.

When it doesn't work, it's always at the same point, just after you press space on a large scroller and before some little animated cow chewing the cud in the middle of the screen.

EDIT: Sorry, I did mention a couple of A500 demos that were glitchy. Well, if I recall, they were:

Virtual Meltdown by Paradise (several little glitches)
End of the Misery by Sagazity (see below)

The trouble is, with the second one, the glitch is gone if I run it in A600 mode, only to crash later on, but in A500 mode the glitch is there but it runs all the way through. That's what I remember.

[Toni Wilen]

Both need ECS Agnus.

[Foebane]

Originally Posted by Toni Wilen

Both need ECS Agnus.

You're right! Just tested them. How did you know? Do you have a tool for analysing disk images or is it just experience?

I forgot them before, but another one that glitches badly is:

Delirium by Complex

And this one has a slight tiny glitch in a picture of an astronaut with a huge gun:

Vector Exterminator by Shining

[Toni Wilen]

This and 3 above posts moved to more correct thread.. (These are not AGA programs..)

Originally Posted by Foebane

You're right! Just tested them. How did you know? Do you have a tool for analysing disk images or is it just experience?

That kind of display corruption (line "shifting" effect) is practically always OCS/ECS incompatibility.

I forgot them before, but another one that glitches badly is:

Delirium by Complex

It is correct behavior! See post #762 above.

And this one has a slight tiny glitch in a picture of an astronaut with a huge gun:

Vector Exterminator by Shining

I am not so sure it is a glitch. Don't assume anything when testing demos..

[SaphirJD]

http://janeway.exotica.org.uk/release.php?id=8970

Not sure if this is really a glitch, i tried backwards until Winuae 3. Anyway, i always missed to capture the glitches in question, so the most close pic is this here (where the part with the text are not "normal" disappearing but instead in a strange way is vanishing, and a bit earlier some strange big glitch at the bottom is visible in Winuae 3.2.2 and 3.3.0 beta's)



Tested with Quickstart A500 1.3 Rom Agnus, 512 KB Chip and 512KB Slow

And also just to be sure with Quickstart A500 1.3 Rom OCS with 512 KB Chip and 512KB Slow - where also another glitch appeared (not sure which one of that 2 configs is the right one for the demo )

[Toni Wilen]

(very slow reply)

Not sure if this is really a glitch, i tried backwards until Winuae 3. Anyway, i always missed to capture the glitches in question, so the most close pic is this here (where the part with the text are not "normal" disappearing but instead in a strange way is vanishing, and a bit earlier some strange big glitch at the bottom is visible in Winuae 3.2.2 and 3.3.0 beta's)

Do you mean the last line before border color? It has random black gaps and flickers on real A500 too.

And also just to be sure with Quickstart A500 1.3 Rom OCS with 512 KB Chip and 512KB Slow - where also another glitch appeared (not sure which one of that 2 configs is the right one for the demo )

It is random sprite garbage, visible only if OCS Agnus. So I guess it is ECS only, at least accidentally

[Toni Wilen]

Smooth Copper looks now correct. Added hack that "fixes" it. Nothing is really emulated but it really is not worth the trouble just for one program but this demo is special enough to do something with it

Technical details:

Demo writes to VHPOSW, changing horizontal counter value exactly by one, this causes copper waits to end 1 cycle earlier (or later) which basically means same behavior as copper using odd cycles to update custom registers. (2 lores pixel resolution, like the demo loader screen says)

Because this also changes bitplanes, it uses BPLCON1 to compensate for bitplane corruption, by scrolling exact same amount to opposite direction of corruption caused by VHPOSW writes.

There is another VHPOSW at the end of each scan line that compensates previous VHPOSW adjustment, syncing Agnus and Denise counters again and preventing hsync problems.

Coding this demo must have taken lots of attempts until timing was exactly right... Even single cycle error would break hsync timing.

[Mclane]

Have to say I love that demo, its a real embodiment of demo's from that era...

[amilo3438]

Megademo by Sargon
http://janeway.exotica.org.uk/release.php?id=2347

Tested with A500 quickstart (most common).

On the 2nd part of demo, some glitches appear below text scroller after cca 1 min.

[Toni Wilen]

Originally Posted by amilo3438

Megademo by Sargon
http://janeway.exotica.org.uk/release.php?id=2347

Tested with A500 quickstart (most common).

On the 2nd part of demo, some glitches appear below text scroller after cca 1 min.

Seems to be demo bug. I booted that demo few times on real A500, sometimes the corruption comes in 10s, sometimes after few minutes. (and again after few more minutes). Emulation not being random (to keep every boot repeatable, as I have said previously) makes the corruption non-random.

[dylandog]

Originally Posted by amilo3438

Nanotech Mysteries 2 by Limited Edition

QS A500 OCS/ECS ... 2 strange vertical lines appear?!

seems fixed
[ Show youtube player ]

[SaphirJD]

Beta 8 introduced some bug for example in Sanity -Arte - rotating pictures/logos look screwed up - was still working in Beta 7 - and looks screwed up also after Beta 8.

https://demozoo.org/productions/5784/

Picture here:



Config is normal A500 quickstart 1.3 OCS config

[Toni Wilen]

Replied in beta thread.

btw, Enchantment under the sea / Resistance interference part has corruption if OCS chipset which is 100% correct behavior. No corruption if ECS.