Glitchy horizontal lines in Vanish's "Monoxide" demo plus list of older reports

[Maren]

As weird as it seems, you're right, that definitely "fixes" it

[Maren]

Sorry, I don't think I'll be able to run any more tests on my miggy (at least not in the near future) since the floppy drive is about to kick the bucket. For a while I blamed it all on the disks which I had re-written a billion times, but the other day I was handed a couple of games which booted ok on another stock A500 but failed miserably on mine.

Anyways, here's a BETA problem but I'm posting it here for reasons previously discussed.

The demo is (once again) "Meeting Collection" by Dope Denmark.


Glitchy


Normal

The CRC32 for the disk is 066EC590. Works ok in 2.0.1 with essentially the same config and all disabling CE does is create even more glitches.

Logs, config and savestate here

Cheers.

[Toni Wilen]

Originally Posted by Maren

Sorry, I don't think I'll be able to run any more tests on my miggy (at least not in the near future) since the floppy drive is about to kick the bucket. For a while I blamed it all on the disks which I had re-written a billion times, but the other day I was handed a couple of games which booted ok on another stock A500 but failed miserably on mine.

Anyways, here's a BETA problem but I'm posting it here for reasons previously discussed.

The demo is (once again) "Meeting Collection" by Dope Denmark.

I can't duplicate (and don't bother with config files, quickstart or nothing)

EDIT: statefile also shows correct graphics.

[Maren]

Default Quickstart config
Default Quickstart config + CE on

Sorry but I think it's either savestates being completely useless or you not waiting enough for the glitch to happen. Try the disk image (no savestate) and once you get to the actual routine, wait a whole minute without taking your eyes off the screen. If you don't get to see any garbage flashing, then maybe it's my computer.

Here's the logs, pure Quickstart this time.

[Toni Wilen]

I saw nothing being said about it happening rarely = I assume it happens instantly. The end.

EDIT: I can see the glitch now. (but I still expect better report next time )

A500 test edit: no above glitch but sometimes whole logo shifts 4 pixels left and later back to original position.

[Maren]

Originally Posted by Toni Wilen

I saw nothing being said about it happening rarely = I assume it happens instantly. The end

Actually it does happen instantly, right after you load the savestate, but it doesn't stay long so you gotta wait till the next flash if you didn't catch it

Anyways, here's another one: "Megademo 2" by Dawn.



This is a garbage frame from several that flash consecutively as the routine "Waky Bobs" starts.

But that's not the problem, there's other garbage frames in the demo, but after you've loaded (and quit, with a left click) Waky Bobs, no other routine (selectable from the main menu) will load again. They all load fine no matter the order, but loading Waky Bobs before anything else will render the rest of the routines unwatchable. Post-waky-bob routines will halt at the Loading/Decrunching screens with music forever looping in the background.

. What you see in the log is me loading the disk, starting Waky Bobs, leaving it and then trying out the first routine which, of course, didn't load.

. The garbage frame will flash about as soon as the savestate is loaded.

. The CRC32 for the .adf is A1DB908A.

. Default Quickstart + CE on was used.

. Disabling CE doesn't help.

. Happens with 2.0.2 B15 as well.

. I recall something similar happening with my A500, but wanted to make sure.

. Get the savestate here

[Toni Wilen]

Originally Posted by Maren

Anyways, here's another one: "Megademo 2" by Dawn.

Exact same behavior on my real A500. Loading-freeze seems to be related to disk motor being off, probably uses some stupid timing that works only with some specific floppy drives.

Meeting Collection difference between working and non-working is 2 cycles, I think problem is not CPU timing but "weird stuff" happening when writing to bitplane custom register when DMA is already active. (real Amiga does that weird shift in this case) Also Copper and CPU writes to custom chipset does not seem to have exact same timing.

[Maren]

Here's a glitchy line occurring at the very beginning (where the savestate was recorded) of the demo "Simplicity" by Associated, contained in the compilation disk "Surprise Pack 04" by Argon. It's that one right above the bottom blue/purple scrolling text.

This happens with the default config, with or without CE on and no other setting seems to make it go away either. Behaves exactly the same in 2.0.2 beta 19.

As usual, not the slightest idea whether this is normal or not, so it's barely another worthless "just in case" report

The CRC32 for the disk is 65817F72.

Get the savestate here

Cheers.

[Toni Wilen]

Originally Posted by Maren

Here's a glitchy line occurring at the very beginning (where the savestate was recorded) of the demo "Simplicity" by Associated, contained in the compilation disk "Surprise Pack 04" by Argon. It's that one right above the bottom blue/purple scrolling text.

A500 confirmed. Probably the usual problem, requires some specific configuration or pack disk maker broke it by recompressing or something like that.

[Maren]

Alright, thanks again for having looked into it

[Maren]

Here's a screenshot from "Madness 2 - The Search for Sanity"'; notice the garbage flashing on the top of the word, right below the red line (which's actually blue but apparently the video codec wanted it to look red)

Both current stable release and beta appear to be affected and disabling CE doesn't help.

The CRC32 for Madness2.exe is 2804502B

Default settings + CE were used to record the savestate, which can be downloaded here

As usual, I'm sorry I can't guarantee this is not reproducible on a real A500, just reporting ugly stuff here and there

Cheers.

[Toni Wilen]

Originally Posted by Maren

Here's a screenshot from "Madness 2 - The Search for Sanity"'; notice the garbage flashing on the top of the word, right below the red line (which's actually blue but apparently the video codec wanted it to look red)

Confirmed. Probably some CPU instruction timing problem again because blits are nothing special.

[Maren]

This one is from Scoopex's "Seven Sins" (CRC32: DA194D64) Zoom in to the white balls and notice those displaced-looking horizontal gray lines with the orange pixels on them. This doesn't happen with CE off and I don't quite remember ever seeing them on my A500, but unfortunately the one I watched was part of a compilation disk so God knows if it's exactly the same exe.

As usual, default config and the latest official release (2.3.0) were used for testing and recording the savestate.

Get the savestate here: http://filesurf.ru/207861/glitch.uss.html

Regards.

P.S. if you want me to quit taking the piss with these false alarms and "nonspecial blits", let me know

[Maren]

This one is from "Magic Demo" by Diabolics. Notice the white line on the checkerboard moving left to right. I'm a 100% certain this never happened on my stock A500, and I'm about as certain that I ran the same exe since I couldn't really find another version online.

Happens with 2.3.0 as well as with 2.3.1. Default config + CE were used to record the savestate, but disabling CE doesn't get rid of the white line anyways.

I must also notice that back when I watched demo on my A500, I still had a hard drive installed so MAYBE there's this slim chance it could have worked "correctly" just because of that (it wouldn't be the first time a peripheral "fixed" a demo) With that in mind, I added a hardfile to replicate my hard drive, placed the demo in it and ran it from there (using my Workbench 1.3 installation) but I still couldn't kill the white line.

Get the savestate here

Get the demo here (CRC32:28BFA093)

Regards.

[Toni Wilen]

It disables sprite DMA just before last sprite used for star field is fetching data.
Same happens on real A500 if sprite DMA is disabled 2 DMA slots earlier, I guess this means some kind of delay needs to be emulated..