WinUAE 4.3.0 beta series

[ross]

Originally Posted by Toni Wilen

Beta 3:

- Subpixel emulation + borderblank: sprite end was 1 hires pixel before bitplane end, leaving 1 hires pixel gap in right border.
- On the fly chipset change support improved. (More OCS/ECS/AGA custom registers differences supported)
- ECS Denise + Subpixel emulation mode + Superhires Display panel setting + not Superhires mode: do not enable ECS Superhires (max 4 colors) emulation mode.
- GUI corruption in some theme modes fixed (b2).

[hexaae]

Originally Posted by Toni Wilen

http://www.winuae.net/files/b/winuae_4300b3.7z
http://www.winuae.net/files/b/winuae64_4300b3.7z

Beta 3:

- Subpixel emulation + borderblank: sprite end was 1 hires pixel before bitplane end, leaving 1 hires pixel gap in right border.
- On the fly chipset change support improved. (More OCS/ECS/AGA custom registers differences supported)
- ECS Denise + Subpixel emulation mode + Superhires Display panel setting + not Superhires mode: do not enable ECS Superhires (max 4 colors) emulation mode.
- GUI corruption in some theme modes fixed (b2).

[amilo3438]

I'am just watching "Freezers-Lech" demo ... seems like some problem after building rotation part in Beta3 ! (quickstart A1200 basic)

EDIT: Seems like same problem in "SCOOPEX-Alien" demo at some parts also! (And in "ANDROMEDA-Nexus7" demo also!)

[ross]

Originally Posted by amilo3438

I'am just watching "Freezers-Lech" demo ... seems like some problem after building rotation part in Beta3 ! (quickstart A1200 basic)

You are right.

Is BRDSPRT=1 ignored after DIWSTOP x position
(so background sprites are variabily blanked at right screen position).

EDIT.
A save state for Toni.

EDIT2: removed

[Toni Wilen]

Fixed. I only tested left side border sprites

[ross]

Originally Posted by Toni Wilen

Fixed. I only tested left side border sprites

Works

[amilo3438]

More problems ...

Found some gfx glitches in "Noxious-BeyondBelief" demo after the "wet-window" part and also in part after that! (not happens in Winuae4210)

Testing config: quickstart A1200 4MB Fast + chipset/cycle-exact (full) enabled

[rsn8887]

How can I test super hires and sub pixel? Which games (if any) support these features?

[Zarnal]

Originally Posted by amilo3438

More problems ...

Found some gfx glitches in "Noxious-BeyondBelief" demo after the "wet-window" part and also in part after that! (not happens in Winuae4210)

Testing config: quickstart A1200 4MB Fast + chipset/cycle-exact (full) enabled

Hi,

same issue.

2 savestates are in the 7z archive.

[Toni Wilen]

"Removed all 68020 cycle-exact CPU mode internal idle cycles. It wasn't good enough design (as expected), need to start from scratch later.. again.. It made real 68020 best case results too slow. (and worst case was still too fast..)"

-> these kinds of "bugs" are normal. As usual, something now works, something works less. All normal.

EDIT: in other words, don't compare full cycle-exact mode differences. If there are differences in memory cycle exact (or without CE), them it could be "real" bug.
Next attempt: perhaps next summer..

[amilo3438]

Originally Posted by Toni Wilen

"Removed all 68020 cycle-exact CPU mode internal idle cycles. It wasn't good enough design (as expected), need to start from scratch later.. again.. It made real 68020 best case results too slow. (and worst case was still too fast..)"

Of course, no use of modeling cycle-exact 68020 CPU mode idle cycles without knowing real info of how 68020 internal sequencer works. (it would be just a guessing, IMHO)
Leave it as it ... it behaves "good enough" in many cases. In other cases, turning off the cycle-exact and playing with "approximate" slider could possibly help.

EDIT: Anyway, it behaves better now than before.

[superfrog]

Beta 3 still has some GUI corruption issues on my setup, though it is still is a huge improvement over Beta 2. You can see from the screenshots that portions of interface elements from the window immediately behind that of WinUAE "show through" when list elements are selected from the left side of the WinUAE GUI.

Clicking elements anywhere else within WinUAE doesn't seem to cause any trouble; it seems that only the list elements on the left have any issues. There are also no apparent problems when there is no window behind WinUAE.

I am running Windows 7 with the Windows Classic theme on an nVidia GT 425M GPU. Startup logs attached.

[DamienD]

I'm sometimes getting sections still highlight in blue when clicking on another:



I clicked "Configurations" first and then when selecting any other sections, "Configurations" is still highlighted.

[Toni Wilen]

http://www.winuae.net/files/b/winuae_4300b4.7z
http://www.winuae.net/files/b/winuae64_4300b4.7z

Beta 4:

- AGA border sprites enabled: sprite was still clipped to horizontal display window end. (b3)
- Fastlane SCSI DMA controller address configuration emulation was incomplete. (Address bus is used for bits 0 to 23, upper 8 data bits for 24 to 31)
- A2024 monitor didn't show full viewable area. (4.0) It still won't work correctly if multimonitor mode is enabled. A2024 is a special case, it is basically display port adapter but it is not hardwired to chipset timing/display size/position like all other display port adapters.
- Remaining (?) GUI update glitch fixed. (b2)
- Added Trojan light gun preset to Game ports panel. Trojan fire button is joystick left. Standard light pen/gun fire button is 3rd mouse/joystick button.
- PPC board without ROM or zero size ROM and 68060 CPU with FPU: Generated fake boot ROM didn't include FPU disable boot code. Since 4.1.0 68060 FPU is not disabled when emulated accelerator board is configured.

[DamienD]

Thank you Master Wilen

[AMIGASYSTEM]

Toni I noticed that on OS 4.1 Emulation active JIT + Video Card Picasso IV, WinUAE crashes, if active JIT + UAE Zorro III Card everything works regularly

I used with WinUAE 4300 beta.x the same config and same HardFile !

[Toni Wilen]

Some lagless beamracing adjustment testing.

Replaced D3DKMT_GETSCANLINE calls with simple "time since last vsync" scanline calculation and at least on my hardware beamracing is again stable (some nvidia driver or something update made it quite unstable some time ago).

D3DKMT_GETSCANLINE seems to require some internal driver lock that can stall driver operations and after whatever update it is even more slow..

http://www.winuae.net/files/b/winuae.7z

Notes:

END+F9 shows color bars that shows stability of timing.
Slice vertical positioning still needs some extra handling.

[DamienD]

Originally Posted by Toni Wilen

Windows 8 or later only (requires D3DKMT_WAITFORVERTICALBLANKEVENT)

...but at the end of the day, beamracing never worked on my Windows 7 [64-bit] / NVidia GeForce GT 540M laptop anyway.

[Toni Wilen]

Originally Posted by DamienD

...but at the end of the day, beamracing never worked on my Windows 7 [64-bit] / NVidia GeForce GT 540M laptop anyway.

Above was wrong, it is supported since Vista. There was side-effect which that made it useless in earlier Windows versions but this method should be compatible.

winuae.7z updated. (but laptops will always have problem with this method..)

Originally Posted by AMIGASYSTEM

Toni I noticed that on OS 4.1 Emulation active JIT + Video Card Picasso IV, WinUAE crashes, if active JIT + UAE Zorro III Card everything works regularly

I used with WinUAE 4300 beta.x the same config and same HardFile !

Not enough information.
- no logs
- no config
- no information which was last working version

[DamienD]

Originally Posted by Toni Wilen

Above was wrong, it is supported since Vista. There was side-effect which that made it useless in earlier Windows versions but this method should be compatible.

winuae.7z updated. (but laptops will always have problem with this method..)

Thanks for the update, might as well give it a try anyway.