WinUAE 4.9.0 beta series (Was 4.5.0)

[BlackByte]

sorry, i have read it: - Removed ROM scan expansion device result list. It is out of date and code was almost unmaintainable.

[Toni Wilen]

Originally Posted by BlackByte

Hi Toni, my OS4.1 configuration crashes immediately when loading. (have tested all betas)

You didn't mention if it works with 4.4?
Also "crashes" equals "it does not work" which are not allowed here. What crashes? WinUAE? Emulated Amiga? Guru? And so on..

[BlackByte]

Yes Toni, with version 4.3, 4.4, there are no problems.
During version 4.9b26, my configuration does not load.
Crashes when loading. It comes the pink picture and end

[Toni Wilen]

https://download.abime.net/winuae/fi...uae_4900b27.7z
https://download.abime.net/winuae/fi...e64_4900b27.7z

Chipset updates are almost done. Optimizations added. (Which might break something that wasn't previously broken)

Beta 27:

- Graphics glitches fixed (b26 and b25)
- Some programs (for example Wings of Death) that use OCS compatible 60Hz hack didn't have working vblank interrupt. If write to VPOSW jumped to mid last line, vblank start line check was missed.
- Added some optimizations (for example don't create new bitplane allocation table for current scanline if previous line had identical bitplane cycle sequence)
- Added bitplane/audio DMA conflict emulation that can happen in bitplane overrun situations. In this situation no DMA transfer is done (neither BPL or audio).
- Integer scale filters usually forced all scanlines to be fully redrawn, even if content didn't change.
- VPOSW fake 60Hz now uses 60Hz height if refresh rate is >=55Hz and 50Hz height if less than 55Hz. Matches 1081/1084 etc CRT behavior better.
- "Remove interlace artifacts" last line flickering fixed.

[hexaae]

Still unoptimal "Remove interlace artifacts" testing Dark Seed (Ross fixed version), with flickerings...

[mark_k]

Originally Posted by Toni Wilen

AFAIK 60HzEmulator uses same hack as Dynablaster on OCS to "switch" to 60Hz. (Too lazy to test)

Yes.

It was probably a tool on the Macrosystem Deinterlace card disk that I used to adjust the number of lines.

In addition to seeing where the height switch occurs, you can see what range of refresh rates your monitor can sync to. A typical 1084 might be something like 47Hz-63Hz. Maybe wider for monitors with a user-accessible vertical hold control.

[hexaae]

Uhm, attached 4 crashes dumps... When running Deluxe Galaga AGA after I ran other WB/games...
https://1drv.ms/u/s!ApMUGr0cuN39gf53...gPaEw?e=QCeIQ1
I can reproduce it somehow if you need and those dumps are not enough, Toni...

EDIT:
can't reproduce with old backup of b16...

[BlackByte]

b27, same problem

[Toni Wilen]

Originally Posted by BlackByte

b27, same problem

This is fixed in next beta. I managed to duplicate it with your config. It had nothing to do with OS 4.1 but "too many" network devices that caused buffer overflow. "Fortunately" my system also had too many and debug build detected the overflow (without it dump file would have been needed to find it). Overflow has always been there but previously it probably overflow to something that didn't cause immediate crash. And probably corrupted something else..

[amilo3438]

Expiration (Demo) by Mayhem - still glitches at cube rotation part but only with the 68000 CPU! (with 68010 it is fine)(A500 OCS/ECS QuickStart config. Beta27)

Not sure why this is happening as in WinUAE4400 was fine with 68000 CPU.

=> NOTE: Also while changing the CPU (on the fly) from 68000 to 68010 on the same demo but "Sepultura" star rotation part the emu just stops! (was fine on WinUAE4400) <=


EDIT: Same for ...
Spiritual connection by Amaze -> http://eab.abime.net/showpost.php?p=...&postcount=101 (as shown on picture)(but here it happens with 68010 too!)

[Toni Wilen]

Originally Posted by amilo3438

Expiration (Demo) by Mayhem - still glitches at cube rotation part but only with the 68000 CPU! (with 68010 it is fine)(A500 OCS/ECS QuickStart config. Beta27)

Not sure why this is happening as in WinUAE4400 was fine with 68000 CPU.

I think this part is half-broken. Sometimes it works, sometimes it does not. Unfortunately it would take too long to test it dozens of time on real hardware. (Similarly broken like that demo that had bouncing ball or something that glitches randomly, sometimes in 10s, sometimes after few minutes, don't remember the name)

EDIT: Same for ...
Spiritual connection by Amaze -> http://eab.abime.net/showpost.php?p=...&postcount=101 (as shown on picture)(but here it happens with 68010 too!)

This changes blitter registers when blit is active. It does not anything too weird so I am not sure whats the real reason. But it is also possible previous "blitter active" fixes aren't anymore fully working, need to do some tests later..

[amilo3438]

Originally Posted by Toni Wilen

(Similarly broken like that demo that had bouncing ball or something that glitches randomly, sometimes in 10s, sometimes after few minutes, don't remember the name)

Hi Freaks (Intro) by Supplex & The Silents

[BlackByte]

@This is fixed in next beta. I managed to duplicate it with your config. It had nothing to do with OS 4.1 but "too many" network devices that caused buffer overflow. "Fortunately" my system also had too many and debug build detected the overflow (without it dump file would have been needed to find it). Overflow has always been there but previously it probably overflow to something that didn't cause immediate crash. And probably corrupted something else..





Thanks Toni

[amilo3438]

Originally Posted by amilo3438

Expiration (Demo) by Mayhem - still glitches at cube rotation part but only with the 68000 CPU! (with 68010 it is fine)(A500 OCS/ECS QuickStart config. Beta27)

Not sure why this is happening as in WinUAE4400 was fine with 68000 CPU.

=> NOTE: Also while changing the CPU (on the fly) from 68000 to 68010 on the same demo but "Sepultura" star rotation part the emu just stops! (was fine on WinUAE4400) <=


EDIT: Same for ...
Spiritual connection by Amaze -> http://eab.abime.net/showpost.php?p=...&postcount=101 (as shown on picture)(but here it happens with 68010 too!)

I forgot to check the last working beta...
So, last working fine was in beta 21 and broken from beta 22! (both demos)

[amilo3438]

Megademo by Intuition

A500 OCS/ECS ... problem on lady part ?! (bitplane shifted?)
It starts from beta20! (last fine in beta19)


Stuff to Love (Intro) by Savage

A500 ECS ... problem on scroller text ?! (shifted in ECS but fine in OCS?) EDIT: ECS A500 confirmed! (see post #839)
It starts from beta18! (last fine in beta17)


Rampage (Trackmo) by The Electronic Knights

A500 OCS/ECS ... just stops before the plane part ?!
It starts from beta22! (last fine in beta21)


Child's Play 1 - Preview (Intro) by Silicon Ltd

A500 ECS ... scroller glitches ?!
It starts from beta27! (last fine in beta26)


Powertrax (Musicdisk) by The Light Circle

A500 OCS/ECS ... weird line appears (as shown on pic.1) ?!
It starts from beta20! (last fine in beta19)

[hexaae]

Marvin's Marvellous Adventure (AGA)

No half-pixel left screen border in b27 (which is fine for me but once we verified it was present on real hw ). b16 still got it.

[Toni Wilen]

Originally Posted by amilo3438

Megademo by Intuition

And probably others happen because I forgot to implement one feature in bitplane pipeline, initial start vs BPLCON0 (resolution/number of planes) change does not have same "delay". I already noticed this problem few weeks ago but I thought it was caused by something else..

Back to drawing board and breaking already working programs more

[Toni Wilen]

Originally Posted by hexaae

Marvin's Marvellous Adventure (AGA)

No half-pixel left screen border in b27 (which is fine for me but once we verified it was present on real hw ). b16 still got it.

Blanking color selection was rewritten and I missed this special case handling. Fixed.

[Toni Wilen]

Originally Posted by amilo3438

Stuff to Love (Intro) by Savage

A500 ECS ... problem on scroller text ?! (shifted in ECS but fine in OCS?)
It starts from beta18! (last fine in beta17)

This is actually correct! ECS A500 confirmed. It uses undocumented feature (switches bitplane DMA off when DMA is is active, result is that when DMA is enabled next time, bitplane DMA continues immediately for one bitplane block) that could not be emulated properly with old code.

[amilo3438]

Originally Posted by Toni Wilen

This is actually correct! ECS A500 confirmed. It uses undocumented feature (switches bitplane DMA off when DMA is is active, result is that when DMA is enabled next time, bitplane DMA continues immediately for one bitplane block) that could not be emulated properly with old code.

Great!

I noticed with this demo the following: When on the fly changing from ECS to OCS it loses the scrolling text! Seems this happens only in the latest b26 & b27! (try more times on b26)