slam tilt intro problem

[amilo3438]

Starting from WinUAE 3500 it jumps over intro part in the Slam Tilt !?
Used config: Quickstart A1200 (basic)

[Foebane]

Provide the links where you got the files for Slam Tilt from, whether it's ADFs or WHDLoad, and the usual logs, of course.

[amilo3438]

Slam Tilt (1996)(21st Century)(AGA)[cr DLM](Disk 1 of 5).adf (CRC = A04CEC7A)

Not sure that log can help in this case!

When it shows the picture from newspaper, after that is should run an intro part, but in WinUAE 3500 and WinUAE 3600 betas it just jump over intro and starts to load... seems fine in WinUAE 3400.

[Foebane]

Originally Posted by amilo3438

Slam Tilt (1996)(21st Century)(AGA)[cr DLM](Disk 1 of 5).adf (CRC = A04CEC7A)

Not sure that log can help in this case!

What website did you get this file from? If I could download it, I could test. Looks like TOSEC filename convention to me.

[amilo3438]

Dont remember, just type the name into google and it might be that.

[Foebane]

Originally Posted by amilo3438

Dont remember, just type the name into google and it might be that.

I found the first result on Planet Emulation, so I tried the first disk in that series there (no [a] or whatever) - it seemed to work fine.

What does the intro of Slam Tilt contain? I've seen the WHDLoad version, and it seems to be just a newspaper cutting that defines what Pinball is, a few credits and a pinball animation, and the menu.

And you're also saying that the ADF worked fine before, but has stopped doing so in the latest WinUAE? I'd say that's definite grounds to provide logs for Toni.

[amilo3438]

Originally Posted by Foebane

What does the intro of Slam Tilt contain? I've seen the WHDLoad version, and it seems to be just a newspaper cutting that defines what Pinball is, a few credits and a pinball animation, and the menu.

Yes, a few credits after newspaper cutting is not shown in WinUAE versions after 3400!

[Rotareneg]

Just checked and 3.4.0 and 3.6.0 beta 10 play that particular cracked version differently: 3.4.0 partially plays the game intro and 3.6.0 skips over it as soon as the music starts. The intro runs in its entirety with 3.6.0 with the uncracked IPF. Here's a video showing the difference between 3.4.0 and 3.6.0 with the crack, and 3.6.0 with the IPF, with loading times edited out:

[ Show youtube player ]

[Rotareneg]

In 3.6.0 beta 10 if you use the A1200 4 meg fast RAM quickstart config the cracked version will get almost all through the intro, skipping at the point where the coin insert sound effect plays, right before the big "SLAM TILT the pinball game" logo comes up.

[amilo3438]

Thanks for checking and confirming!

EDIT: Now is left that someone check it on a real A1200! (basic configuration)
If it happens the same no problem, if not than the problem exist.

[Tomislav]

WinUAE 3.4.0 and 3.4.0 b20 are ok

Problem starts with 3.4.1 b1. There are many changes in that b1 version: http://eab.abime.net/showthread.php?t=85939

When I uncheck "Cycle-exact (Full)" intro runs.

(There was not 3.4.1 stable because it ends with 3.4.1 b9 and starts with 3.5.0 b10, then next stable is 3.5.0.)

[amilo3438]

It is not a problem if it happens also on an real A1200 non-expanded, but this needs to be confirmed first by someone else. (thx for testing)

[Toni Wilen]

It does not happen on real A1200 but reason is still the same as always with 68020 so there is nothing new.

[amilo3438]

Interesting is that the .ipf version (see post #8) runs fine on the emu, and only the crack version is problematic!
Also, does we know how it behaves on a real machine when the cache is disabled?

[Galahad/FLT]

Originally Posted by amilo3438

Interesting is that the .ipf version (see post #8) runs fine on the emu, and only the crack version is problematic!
Also, does we know how it behaves on a real machine when the cache is disabled?

Have you tried running without the cracktro just on the off chance it does something?

[Toni Wilen]

I didn't examine it much but it looks like code does something that takes a bit too long in emulation (if "cycle-exact") and flag that means something like "exit requested" is not cleared quickly enough and it is checked in vblank while main routine code hasn't reached the clear part yet.

Or something like that..

EDIT: It may be even as simple as crack changing timing slightly (time until next vblank), triggering bug in original code which didn't clear the flag early enough.

[amilo3438]

If you enter into debugger, you can see that the cracked version runs with CACR=0 where the uncracked runs with CACR=1. If you change on the cracked version in the debugger to CACR=1 it will run fine as uncracked version.
So, if uncracked version works fine on a real A1200 (unexpanded) that could mean the emulation with cache turned OFF is maybe too slow.

Edit: In fact, for some still problematic demos is vice-versa, i.e. works "fine" (but too slow) only if cache is disabled.

Cheers!