Problematic Demos.

[Toni Wilen]

Originally Posted by Saghalie

Working with 2.8.1 see attached.

I didn't try to get it working with the quick start. I just kept ticking switches until it worked.

I think you missed the point of this thread..

[Saghalie]

oh...

[amilo3438]

Originally Posted by mark_k

Yep, it first breaks (graphical corruption then crash at rotating/warping purple Old Bulls image) with WinUAE 2.7.1b1, so maybe one of the many changes for better emulation accuracy wasn't quite right.

It first breaks with test winuae_2710b0.exe (2013.12.22. 17:20) version.

I still have all those test versions from my active period of demo testing.

Mention it only for historical correctness. ;-)

Btw. quick and dirty fix for that problem would be to enable Immediate Blitter in Chipset options.

[vagrant]

Great
Yeh I noticed this one after a request to capture it for youtube.
Will report if I come across any others

[Toni Wilen]

Last minute update, I decided I had to fix this after all

http://www.winuae.net/files/b/winuae.zip

It was cycle sequence bug when copper wakes up and was waiting for blitter, it was as "simple" as copper should wait for blitter first, then normal copper WAIT (was the other way previously). This moves WAIT idle cycle to correct position, result is wakeup cycle sequence that matches real hardware without 271b1 update (which was not correct, it only looked like it was correct).

[amilo3438]

Skip Intro (Intro) by Awesome & D-Tect & Vanish -> http://janeway.exotica.org.uk/release.php?id=6980

A500 OCS (most common) problem in OCS!

winuae 2.7.1beta4 (2014.01.18) OK in OCS!
winuae 2.7.1beta5 (2014.02.02) broken in OCS!

p.s. The demo was released a few days ago on Kestra BitWorld demobase! (and test it just a few minutes ago!)

[Toni Wilen]

Originally Posted by amilo3438

Skip Intro (Intro) by Awesome & D-Tect & Vanish -> http://janeway.exotica.org.uk/release.php?id=6980

A500 OCS (most common) problem in OCS!

winuae 2.7.1beta4 (2014.01.18) OK in OCS!
winuae 2.7.1beta5 (2014.02.02) broken in OCS!

p.s. The demo was released a few days ago on Kestra BitWorld demobase! (and test it just a few minutes ago!)

It is 99.999% demo bug if OCS has sprite problems 2710b5+. (It is first version that emulated this undocumented OCS/ECS sprite fetch behavior with overscan)

Not going to switch back to OCS Agnus today, don't want to break Agnus socket too soon

EDIT: A1000 tested, confirmed but corruption is a bit more colorful (flickering yellow color also appears in corrupted part) which means SPRxDATB can get changed (with something unknown) even when DMA is inhibited due to bitplane DMA.

EDIT2: A500 OCS (8371) tested, too strange to not test: same glitch but yellow (DATB contents) part is slightly different than on A1000. Most likely difference is Denise, my test A500 has ECS Denise. (Denise contains sprite data registers), not Agnus. Anyway, not that important because this can be only used to create garbage

[amilo3438]

Originally Posted by Toni Wilen

Last minute update, I decided I had to fix this after all

http://www.winuae.net/files/b/winuae.zip

Seems to work fine now! But not going to test today if the fix break something else. (it is too boring)

EDIT: I've already tested some 50 or so new released demos from Kestra... and that would be enough for today.

[amilo3438]

I was bored...

http://eab.abime.net/showpost.php?p=...&postcount=592

seems the above bug is back again from winuae_2900b23.exe (2014-11-01)

[Toni Wilen]

winuae.zip updated.

Second last minute change: A1000 vs A500 last line sprite behavior difference found and implemented. Now graphics glitches (if OCS) in Skitro by Awesome/D-Tect/Vanish are 100% real hardware matching in A500 OCS and A1000 Agnus configurations.

(and quick workaround for Samplers demo)

[amilo3438]

Profecy-LetsRave -> http://janeway.exotica.org.uk/release.php?id=5278

tested in A500 ECS Agnus 0.5MB + 0.5MB... problem after cooper bars part!

winuae_3000b27.exe (OK)
test winuae_3000b27_2014-12-13.exe (broken)

EDIT: enabled Immediate Blitter in Chipset options will fix it!

[Toni Wilen]

Thanks, and copper blitter wait was a bit too early now.. (Very buggy demo but it works by luck..)

Hello!
note not working demo "MadeInCroatia" in WinUAE 3.0.0 (Public Beta 27)
thank you.

[amilo3438]

Originally Posted by Sergey

Hello!
note not working demo "MadeInCroatia" in WinUAE 3.0.0 (Public Beta 27)
thank you.

Thanks for reporting!

broken in winuae2700beta16 (2013.11.30.)

p.s. I missed it because I forgot to put it in my demo test-demobase. (obviously, no one is perfect)

(Seems like a CPU problem... if cycle-exact = OFF it will work!) or maybe something else (probably I was already a little tired)

[Toni Wilen]

Worst ever report but untick "CIA TOD Bug" in Advanced chipset as a workaround.

[amilo3438]

Originally Posted by Toni Wilen

Worst ever report but untick "CIA TOD Bug" in Advanced chipset as a workaround.

yeah, right... I assumed there must be some reason why I do not have it in my test demo base!

Also, I did not notice anything unusual over, at least for those demo mentioned on the first page of this thread. (so I end up with demo testing)

I hope that this will be the best version so far of winuae.

Cheers!

[amilo3438]

Elan Performer 1.0 (Crack Intro) by Creator -> http://janeway.exotica.org.uk/release.php?id=73845

forgot to mention this yesterday while was testing new demos... the problem happens only in ECS Agnus.
(not sure is it a problem or not, or it is normal for ECS)

EDIT: Be careful while switching back to ECS Agnus! (to not break your Agnus socket too soon)

[Toni Wilen]

Originally Posted by amilo3438

Elan Performer 1.0 (Crack Intro) by Creator -> http://janeway.exotica.org.uk/release.php?id=73845

forgot to mention this yesterday while was testing new demos... the problem happens only in ECS Agnus.
(not sure is it a problem or not, or it is normal for ECS)

EDIT: Be careful while switching back to ECS Agnus! (to not break your Agnus socket too soon)

A500 confirmed (I have another ECS A500). This kind of effect is the most common OCS vs ECS compatibility problem.

[amilo3438]

Originally Posted by Toni Wilen

untick "CIA TOD Bug" in Advanced chipset as a workaround.

For me this is new information!

At the time when I tested the demo "MadeInCroatia", bug that I reported was regarding some graphical errors, and WinUAE had by default in the Advanced Chipset the "CIA TOD Bug" set to OFF... and in later versions is become again set to ON by default.

To me at the time was only known that for eg. the demo "The End by trilobite" has problems if this option is not set to ON!

Now several questions bothering me:

How likely each A500 has the "CIA TOD bug", whether listed demo "MadeInCroatia" and "The End by trilobite" have problems when run on a real Amiga 500?

Is there a list of program code that is need to have that option turned off to work properly in WinUAE? (Can such a list be inserted into WinUAE doc. file?)

Would it not be possible on the basis of these lists in detail to study / analyze this problem? (ie. why is the WinUAE at all necessary to have that option when the actual Amiga do not need it)

Thanks.

[Toni Wilen]

CIA TOD bug is the only option that is a sort of a hack fix. It is proven that "TOD bug" exists but for some reason it does not seem to trigger any interrupts in all hardware configurations.

It seems to be only a side-effect in MadeInCroatia, possibly some disk operation may a bit too slow (like
"disk rotating at all" speed state) and there is just enough time to trigger it. Later..

EDIT: MadeInCroatia for some reason does not immediately disable CIA-B interrupts when it takes over the system but only later and if TOD interrupt triggers during that period: crash.