WinUAE 4.9.0 beta series (Was 4.5.0)

[Toni Wilen]

Other listed demos should also work now. (It was copper wait timing problem, not bitplane sequencer)

On the fly switching is known to have some issues, for example it causes annoying flicker.

[amilo3438]

Coffee by Y-Crew -> https://www.pouet.net/prod.php?which=71975

Glitches during the coffie cup transfer from right to left side! (from b25)

Originally Posted by Toni Wilen

Other listed demos should also work now. (It was copper wait timing problem, not bitplane sequencer)

Yes, except above mentioned, other demos work now in this winuae.7z test version!
(Good to here it was not problem with bitplane sequencer.)

Expiration (Demo) by Mayhem - still glitches (fine in b21), but about this is already mentioned in post #831. (if this is correct)

EDIT: Seems "EyeQlazer" and "Hotbleeps" are broken in this winuae.7z test version ?!

Originally Posted by Toni Wilen

On the fly switching is known to have some issues, for example it causes annoying flicker.

This did not happen in some previous betas I mentioned (see underline text in post #834 and #840), so something strange is going on?!


BTW
I noticed on some emu sites at every new beta WinUAE release is mentioned following text: "WARNING: don’t use it you aren’t sure, this is « more beta » than usually."
Is this WARNING still apply for new betas?

[Toni Wilen]

Originally Posted by amilo3438

[B]EDIT: Seems "EyeQlazer" and "Hotbleeps" are broken in this winuae.7z test version ?!

Probably side-effect to previous copper update. This needs more testing, after next beta.

BTW
I noticed on some emu sites at every new beta WinUAE release is mentioned following text: "WARNING: don’t use it you aren’t sure, this is « more beta » than usually."
Is this WARNING still apply for new betas?

I think they are sort of normal betas since last few weeks

[Toni Wilen]

https://download.abime.net/winuae/fi...uae_4900b28.7z
https://download.abime.net/winuae/fi...e64_4900b28.7z

Beta 28:

- Fixed buffer overflow if more than 9 devices were found when enumerating NPCap/WinCap devices. (Special SLIRP cases ignore max size check)
- Mid screen HBlank change didn't force current scan line redraw. It could have caused glitches in some weird situations.
- Fixed crash when display DMA start/stop/position was too weird. (Can happen if setup changes mid scanline with DMA already active)
- AGA borderblank hires pixel delay is working again.
- b27 bitplane allocation optimization fix.
- Copper WAIT can't wake up at the start of scanline. (This was lost when custom emulation rewrite started)
- Sprite emulation fix, vblank end sprite reset and first possible sprite DMA didn't use same line.
- Blitter register modification while active support was partially broken few betas ago.
- ECS Denise H/V-blank emulation updates.
- RTS/RTE/RTR/RTD disassembly shows also return address, calculated using current stack pointer value.
- DMA debugger now includes DDFSTRT (0), DDFSTOP (1) and hardwired DDFSTOP (2) positions if match caused bitplane DMA to start (DDFSTRT) or stop (DDFSTRT/hardwired DDFSTOP).

[amilo3438]

Planet Rocklobster (Demo) by Oxyron

[A500 ECS 512+512] Log window shows: "possible buggy copper cycle conflict with blitter" (close to the end part of the demo)


Oh Five! (Demo) by Pacif!c

[A500 ECS 512+512] Log window shows: "possible buggy copper cycle conflict with blitter" (at the end of part before sphera)

[BlackByte]

Hi Toni, with the current beta28 I have the same crashes with OS4.1 as before.

[Toni Wilen]

Originally Posted by BlackByte

Hi Toni, with the current beta28 I have the same crashes with OS4.1 as before.

Then the cause is something else. Does it really happen immediately when clicking config load button? Does it create crash dump file? You could also load the config file in text editor, then remove parts of config file (for example binary search, remove half of config file, if it still crashes, halve the remaing part etc.., if it didn't crash, restore removed half and halve it. And so on.) until you find the config entry that causes the crash.

[BlackByte]

No, it only appears after loading, with the pink picture. Unfortunately, the dump file is not created.

[Toni Wilen]

Next test: run in windowed mode with -log -logflush command line parameters. Run also 4.4.0 using same parameters (you can quit emulation after pink screen).

Attach both log files (winuelog.txt. winuaebootlog.txt is not needed). Hopefully they contain enough clues..

[BlackByte]

Here are the two files, maybe you can find the cause.

[hexaae]

b28 freezed while I was working with PPaint6.4c, then I opened the GUI, selected Native/RTG On-Screen display to see activity, and it crashed once I clicked OK button in the GUI.
Dump: https://1drv.ms/u/s!ApMUGr0cuN39gf8B...MEEcg?e=wpB5VC

[Toni Wilen]

Originally Posted by hexaae

b28 freezed while I was working with PPaint6.4c, then I opened the GUI, selected Native/RTG On-Screen display to see activity, and it crashed once I clicked OK button in the GUI.
Dump: https://1drv.ms/u/s!ApMUGr0cuN39gf8B...MEEcg?e=wpB5VC

Something overwrote memory that shouldn't be overwritten. Unfortunately it is impossible to know what caused it.

Can you repeat it?

Originally Posted by BlackByte

Here are the two files, maybe you can find the cause.

Some hints but not the cause. Some tests:
- disable UAE boot ROM (ROM panel, set it to disabled)
- disable UAE RTG
- remove PCI bridge

Does it still crash?

btw, set Chip RAM to 2M, OS4 won't like impossible configurations.

[hexaae]

Originally Posted by Toni Wilen

Something overwrote memory that shouldn't be overwritten. Unfortunately it is impossible to know what caused it.

Can you repeat it?

Unlikely, but I'll try...

[BlackByte]

Hello Toni, the Emu also crashes with these settings. I have no idea what that could be.
Version 4.3 and 4.4 runs perfectly

I have set all of your recommended settings
hmm..............

[amilo3438]

Going through this reminder (that usually I forget to check) http://eab.abime.net/showpost.php?p=911206&postcount=59 just found another few demos that still glitches:

Demo 1 (Demo) by The Link > http://janeway.exotica.org.uk/release.php?id=1141

New Demo (Demo) by British Swapping Syndicate -> http://janeway.exotica.org.uk/release.php?id=27934

Tested config: A500 KS1.2 OCS/ECS 512 on WinUAE beta28.

[Toni Wilen]

Copper issue should be fixed now (Hotbleeps etc) and sprite glitch in "ultrademo".

Originally Posted by amilo3438

Log window shows: "possible buggy copper cycle conflict with blitter" (close to the end part of the demo)

It is normal. (or do you mean it happens differently now?)

Originally Posted by BlackByte

Hello Toni, the Emu also crashes with these settings. I have no idea what that could be.
Version 4.3 and 4.4 runs perfectly

I have set all of your recommended settings
hmm..............

I can't duplicate the crash with your config. I even run it with "Address Sanitizer" debugging mode enabled (array bounds check etc) which makes it run really slowly. Nothing interesting happened, it booted normally. Except harmless single byte out of bounds read in uaegfx routine (which happens when OS4 WB appears, your crash is much earlier)

Try creating new config from scratch. If you have default.uae, move it from configurations first. Then run winuae, manually create new config (don't load any configs as a base config). Keep it as minimal as possible, do not enable any expansions that are not required to boot OS4.

Does it still crash?

There is still possibility the real difference is in your OS4 HDF or CSPPC flash settings.

[Zarnal]

Originally Posted by Toni Wilen

Copper issue should be fixed now (Hotbleeps etc) and sprite glitch in "ultrademo".

I just retested, EyeQlazer and Hotbleeps are Ok with official b28 but not with today's version ( 20H26 ).


Ultrademo is now Ok.

[Toni Wilen]

Originally Posted by Zarnal

I just retested, EyeQlazer and Hotbleeps are Ok with official b28 but not with today's version ( 20H26 ).

No, b28 was wrong. Don't trust every screenshot.
(EDIT: diagonal lines = correct. Vertical only = wrong)

[Zarnal]

It's understood.

[amilo3438]

Originally Posted by Toni Wilen

Copper issue should be fixed now (Hotbleeps etc) and sprite glitch in "ultrademo".

Critical Mass by Parallax (ADF from arabuusimiehet) has problem with this winuae.7z!
It stops with loading but the music is still playing! (was fine in official beta28)

Tested config: quickstart A500 1.3 1Mb Chip RAM


EDIT: Rampage (Trackmo) by The Electronic Knights -> ... just stops before the plane part ?! (again) Strange, now works! Not sure what happens.
NOTE: Hmm, my default WinUAE folder (exe dir) has some .ini and .cache files that I now deleted.
I have disabled now in WinUAE path option: cache configuration files, use reletive paths and portable mode. (this was enabled before)
I know that is a lot of variables, but now will use this setting to check if that rarely (random) strange (hardly repeatable) behaving is gone! (so far so good)

Originally Posted by amilo3438

Expiration (Demo) by Mayhem - still glitches (fine in b21), but about this is already mentioned in post #831. (if this is correct)

It seems this behaving is correct! I manage to get it randomly work w/o glitching in this winuae.7z test version!