Problematic AGA demos.

[Zarnal]

@Amilo3438

Oops, this is one of my very old report ( 2018 ).

it's CPU related as usual.

Note : video of this demo in real hardware ( created during our tests on real A1200/WinUae on a French forum ).

[ Show youtube player ]

[amilo3438]

Beut (64k Intro) by Phase Truce --> http://janeway.exotica.org.uk/release.php?id=18554

Last worked fine in WinUAE4300! (Broken from WinUAE4400!)

WinUAE QS A1200 basic config!

[ross]

Originally Posted by Zarnal

A new test on Hard should be done this weekend (maybe) to answer your question. I read your explanation carefully.
You say that WinUae does not emulate this " effect ". Is it because it's impossible to do it ?

WinUAE could somehow emulate this, but probably only as an "Advanced Chipset feature".
At each power up memory chips has these different (in some cases very different) bit patterns, so the demo is visually different on each machine.

Toni would be forced to define a bit-pattern and I don't know how much sense it makes, considering that they are quite random (even if they follow usually a 0/1 'chessboard') ..

Into the Zone! a 'proper' version, using a memory 'chessboard bit-pattern' synthetic fix.

For the interested this is how the patterns are made (d6=largest chip-mem available space):

		moveq	#0,d0
		moveq	#-1,d1

		divu.w	#(2*48*5*8),d6
		subq.w	#1,d6
.3		moveq	#2-1,d5
.2		moveq	#48-1,d4
.1		moveq	#5-1,d3
.0		move.l	d0,(a0)+
		move.l	d1,(a0)+
		dbra	d3,.0
		dbra	d4,.1
		exg	d0,d1
		dbra	d5,.2
		dbra	d6,.3

[Zarnal]

Originally Posted by ross

WinUAE could somehow emulate this, but probably only as an "Advanced Chipset feature".
At each power up memory chips has these different (in some cases very different) bit patterns, so the demo is visually different on each machine.

Toni would be forced to define a bit-pattern and I don't know how much sense it makes, considering that they are quite random (even if they follow usually a 0/1 'chessboard') ..

Into the Zone! a 'proper' version, using a memory 'chessboard bit-pattern' synthetic fix.

For the interested this is how the patterns are made (d6=largest chip-mem available space):

        moveq    #0,d0
        moveq    #-1,d1

        divu.w    #(2*48*5*8),d6
        subq.w    #1,d6
.3        moveq    #2-1,d5
.2        moveq    #48-1,d4
.1        moveq    #5-1,d3
.0        move.l    d0,(a0)+
        move.l    d1,(a0)+
        dbra    d3,.0
        dbra    d4,.1
        exg    d0,d1
        dbra    d5,.2
        dbra    d6,.3

Thank you.

[Zarnal]

Originally Posted by amilo3438

Beut (64k Intro) by Phase Truce --> http://janeway.exotica.org.uk/release.php?id=18554

Last worked fine in WinUAE4300! (Broken from WinUAE4400!)

WinUAE QS A1200 basic config!

Confirmed.


It Freezes during red cube scene.

[Toni Wilen]

Originally Posted by ross

WinUAE could somehow emulate this, but probably only as an "Advanced Chipset feature".
At each power up memory chips has these different (in some cases very different) bit patterns, so the demo is visually different on each machine.

I have mentioned it few times previously that I won't add it until someone explains RAM chips internal layout because thats the important bit (or does dozens of experiments to find out the pattern). Same rows and/or columns seem to have same bit (mostly) at powerup which explains the "chessboard"-like pattern.

[heavy]

Originally Posted by Zarnal

Thank you also for your Fix of the Turnips demo.

your welcome...

[ross]

Originally Posted by heavy

your welcome...

I think he thanked for the upload, no need to smash your head against the wall

To clarify, the thanks for the fix go to someone else.
If it's you, thanks to you!

[heavy]

zarnal will understand...
http://eab.abime.net/showpost.php?p=...&postcount=783

it's just a temp quick and dirty graphic fix because unfortunately it's not a code bug.
If a "final" version existed, it would be better

[ross]

Originally Posted by heavy

zarnal will understand...
http://eab.abime.net/showpost.php?p=...&postcount=783

it's just a temp quick and dirty graphic fix because unfortunately it's not a code bug.
If a "final" version existed, it would be better

ah sorry, I had no idea it was something already discussed in 2019, I follow this thread occasionally and even if I read it at the time I totally forgot about it

[heavy]

@ross: by the way, thanks for the fix of Ubergirl.
yep, another "old" thread about it somewhere on this forum
http://eab.abime.net/showthread.php?p=1420396

[Zarnal]

Originally Posted by heavy

your welcome...

it was a stupid mistake (my bad report for this demo) which wasted time unnecessarily.

[Foebane]

Grid 2 by Nature https://www.pouet.net/prod.php?which=2804
Zeon by Nature https://www.pouet.net/prod.php?which=2803

Originally Posted by Toni Wilen

Seems to work if "No flags" JIT option is unticked (which disables JIT CPU flags optimization, some performance is lost but JIT is still active). Some kind of rare JIT bug, probably difficult to find as usual.

UPDATE:

Finally got these two to work with no problems, by simply turning JIT off (cachesize=0) and using Fastest Possible, adjusting the sound buffer and also turning off Vsync too.

I thought I'd tried it already, but turns out that wasn't the case.

So now that's 0 demos that I need keys to control JIT for, meaning I can get rid of them.

As for Fastest Possible, I wrote it off too soon, as it does come in useful for those demos that are too slow on Approximate A500/A1200 or Cycle Exact.

Thanks for the help, Toni, and sorry for the inconvenience.

[Zarnal]

Some checks were made on A1200 this weekend, some results are very "surprising".

[Zarnal]

WinUae beta 4.9 b31:

I noticed a new problem with the "Noxious-Beyond Relief" demo which seems to be related to the chipset. The background image is distorted sometimes on the beach scene (pic 1). Since b22.

Quickstart A1200.

[amilo3438]

@Zarnal

Bug / glitch tracking WinUae / FPGA / A1200 stock

So there is another place for bug tracking! (this information is on a need-to-know basis)

Note: The point is not to fix a demo to work with the emulator but to fix the emulator to work with a demo!

Btw. Polka Brothers-Twisted should work fine with KS3.1 on the emu - at least worked on an older versions A1200+4Mb. (and crashed with KS 3.0, if I remember correctly)
EDIT: Tested to work on latest beta on A1200 basic CE (DMA/memory) enabled and KS3.0!

[amilo3438]

Benoit (Demo) by Ramses (found by my "counterpart" on AmigaFrance)

[A1200 basic] Got an message: Sorry ! This demo needs an A1200 68020 14 Mhz !

(JFI: There is an version on previous post fixed to work with the emu! ... still glitches!)

[Zarnal]

@amilo3438

This list is only valid with WinUae 4.1. It was the last version when I started.

For the majority of issues of my list, these are CPU concerns. It is for this reason that I have not reported any bugs here. ( example: Ramses-Benoit patched version works fine in CPU fastest mode without black bars ). Another example: Union-Zoombi has unconfirmed glitches on a real stock machine, but on a real machine accelerated with an ACA 68030, the glitches are identical to those of WinUae.

All problematic demos were tested on real stock hardware.

With the new Agnus emulation, we got some demos in cycle mode (FD-Methylate, Passion-Alien inspiration, Passion-Comadose, PK-Planet M and some others). I will retest my entire list from A to Z after the official release of WinUae 4.9.

[amilo3438]

I don't know, I am slowly losing interest for AGA and 68020 because of all still unknown details needed for precise emulation. For now I stay focused on OCS/ECS machines.

Originally Posted by Zarnal

I will retest my entire list from A to Z after the official release of WinUae 4.9.

Hmm, maybe better wait for WinUAE 5.x! (Toni somewhere mentioned something about new findings, so this could probably wait until all other thing is finished, I guess.)

[Zarnal]

Toni, ignore my stupid report on the Noxious-Beyond Relief demo from today. After seeing the same kind of glitches in other demos, I realized that CE full chipset mode was disabled in Quickstart A1200 mode ( ( Why exactly ? i don't know really ). A regrettable inatention. Everything is now back to normal.

I will have to redo all my tests done today later.

Originally Posted by amilo3438

I don't know, I am slowly losing interest for AGA and 68020 because of all still unknown details needed for precise emulation. For now I stay focused on OCS/ECS machines.


Hmm, maybe better wait for WinUAE 5.x! (Toni somewhere mentioned something about new findings, so this could probably wait until all other thing is finished, I guess.)

I am also testing OCS at the moment.