WinUAE 4.9.0 beta series (Was 4.5.0)

[AMIGASYSTEM]

Originally Posted by AC/DC HACKER!

Workbench is now funky colors if I have CGX monitors enabled. Also, if check Warp3D games..and powerpc.library crashes, I'm able to reset (CTRL+LAmiga+RAmiga) and it's not rebooting into a gray or black screen.

The colours look like Shogo did when I reported about that. I'm glad I don't have to exit WinUAE and run it again to use WinUAE now...with this update; after checking 3D games. Cool improvements. (Still Checking..)

With the new fix you should no longer have this problem ?

[Toni Wilen]

When reporting graphics board related issues, include following:

a) Picasso96 or CGX and which version. (There for example internal differences between latest Aminet P96 and recent new versions)
b) Used graphics board.
c) Does it happen with another board. (for example PicassoIV vs CV64/3D)

[Zilog]

Originally Posted by Toni Wilen

When reporting graphics board related issues, include following:

a) Picasso96 or CGX and which version. (There for example internal differences between latest Aminet P96 and recent new versions)
b) Used graphics board.
c) Does it happen with another board. (for example PicassoIV vs CV64/3D)

Hi Toni,
a) write in the name of attachment

b) Graphics card cbv 64 /3d (in the last log the was picassio bios , because when change graphics card , picasso bios remain select. Correct in this report.)

c) With picasso driver i have oher error , " no mode id fund", but i read in internet the s3 virge accelarate,work only 640x480x15bit and in picasso driver i can set only 640x480x16bit, i would not like the error maybe caused by this.

Warp3D v4.2a (aminet)
Program for test file "warptest of package Warp3d

I remain available for other tests.


Cybergraphics_v4.3 rc6.zip
Picasso96_v2.0 (aminet).zip

[Toni Wilen]

c) With picasso driver i have oher error , " no mode id fund", but i read in internet the s3 virge accelarate,work only 640x480x15bit and in picasso driver i can set only 640x480x16bit, i would not like the error maybe caused by this.

Check sys:prefs/env-archive/picasso96/showmodes Make sure it contains single string "All" without quotes. Any other will filter out some modes.

Warp3D v4.2a (aminet)
Program for test file "warptest of package Warp3d

"Unable to open screen. Reason: Error code 3"
"Closing down..."

Really helpful..

EDIT: Under P96 screen opens and then it crashes because of jump to random ROM address.

-> I'll ignore these results unless there is HDF from real hardware with CV64/3D config that is confirmed working.

[Zilog]

Originally Posted by Toni Wilen

.............

-> I'll ignore these results unless there is HDF from real hardware with CV64/3D config that is confirmed working.

Ok Toni,
i thinks is impossbile have a hdf from real AMiGA.
Stop with test how want you.

[amiga_amigo]

Hi Toni,

i have an issue with the 4500 betas (all beta 1-4):

Bridgeboard vga (both CL-GD5426 and CL-GD5429) seems to have a problem with mode switching(?) - at least in windowed mode, haven't used fullscreen.
Starting any game e.g. WingCommander from the dos-prompt, the screen goes/stays black.
I can hear the sound playing. Switching to the "amiga monitor" with <END>+<F9> and back makes the pc vga to show correctly.
No problems afterwards. It doesn't matter if the pc vga is on virtual monitor 1 (i.e. single window, switching with <END>+<F9>) or on monitor 2.
My config and drivers worked fine in 4400 and before.

Greetings, Michael

[Mad-Matt]

Beta4 and Beta4fix keeps resulting in HPos hang/crash/error that jumps up in ever greater number the more you try to close the window. Really annoying to have to dig into task manager to quite uae.

Just a few seconds in a 3d prog seems to trigger it. No such issue encountered in Beta3.

[Toni Wilen]

See above. Because this is emulation getting confused, "only" HDF+config is required. No HDF from real hardware needed.

(This is why I really hate debugging programs that don't take over the system. There are millions of variables and some combinations never trigger it)

EDIT: My guess is there is memory corruption/overflow in virge 3d emulation that overwrites UAE internal variables.

[Toni Wilen]

Originally Posted by amiga_amigo

Hi Toni,

i have an issue with the 4500 betas (all beta 1-4):

Bridgeboard vga (both CL-GD5426 and CL-GD5429) seems to have a problem with mode switching(?) - at least in windowed mode, haven't used fullscreen.
Starting any game e.g. WingCommander from the dos-prompt, the screen goes/stays black.
I can hear the sound playing. Switching to the "amiga monitor" with <END>+<F9> and back makes the pc vga to show correctly.
No problems afterwards. It doesn't matter if the pc vga is on virtual monitor 1 (i.e. single window, switching with <END>+<F9>) or on monitor 2.
My config and drivers worked fine in 4400 and before.

Greetings, Michael

Does having WB in native (and no RTG enabled) or RTG mode have any effect on this VGA switching behavior?

[Zilog]

Originally Posted by Toni Wilen

Check sysrefs/env-archive/picasso96/showmodes Make sure it contains single string "All" without quotes. Any other will filter out some modes.



"Unable to open screen. Reason: Error code 3"
"Closing down..."

Really helpful..

EDIT: Under P96 screen opens and then it crashes because of jump to random ROM address.

-> I'll ignore these results unless there is HDF from real hardware with CV64/3D config that is confirmed working.

Ok. With 15 bit resolution Picasso96 and Warp test is ok.

[amiga_amigo]

Originally Posted by Toni Wilen

Does having WB in native (and no RTG enabled) or RTG mode have any effect on this VGA switching behavior?

Amiga side is just a plain A2000/OCS config.
The only RTG in this system is the bridgeboard vga.
I have no hdf with a RTG-enabled workbench to test with - sorry.

[AMIGASYSTEM]

Toni did some testing with CV3D + Picasso96 and Warzp3D 800x600x24Bit, almost all demos seem to work well, there is not much stability.

Even with CV643D + CyberGraphX and Warzp3D most demos work but it is slower than Picasso96.

Made very small exhaustive video CV643D + Picasso96 800x600x24Bit and Warzp3D:

http://eab.abime.net/zone/Test-CV64-Warzp3D.zip

Screenshot CV643D + CyberGraphX 800x600x16Bit and Warzp3

[Toni Wilen]

https://download.abime.net/winuae/fi...nuae_4500b5.7z
https://download.abime.net/winuae/fi...ae64_4500b5.7z

Beta 5:

- If ADF started with 4 first bytes of executable (0x000003f3), it was mounted as a bootable executable. Even if it was disk image and too large to fit in 880k image. Exe mounting is now only done if size is smaller than 880k and file extension is not known disk image extension. (B.A.T. II disk #4)
- Read random data also if disk image is inserted, drive is selected but motor is stopped. Previously random data was only returned if drive was empty.
- CV64/3D 15/16/24-bit graphics glitch fix (b4 debugging code)
- CV64/3D emulated chip model was wrong, fixes WarpTest in perspective correct mode.

CV64/3D + Picasso96 + Warp3D stuff sort of works. CV64/3D + CGX4 + Warp3D does not, reason unknown. (at least in my test)

[Crashdisk]

Originally Posted by Toni Wilen

https://download.abime.net/winuae/fi...nuae_4500b4.7z
https://download.abime.net/winuae/fi...ae64_4500b4.7z

Beta 5:

- If ADF started with 4 first bytes of executable (0x000003f3), it was mounted as a bootable executable. Even if it was disk image and too large to fit in 880k image. Exe mounting is now only done if size is smaller than 880k and file extension is not known disk image extension. (B.A.T. II disk #4)
- Read random data also if disk image is inserted, drive is selected but motor is stopped. Previously random data was only returned if drive was empty.
- CV64/3D 15/16/24-bit graphics glitch fix (b4 debugging code)
- CV64/3D emulated chip model was wrong, fixes WarpTest in perspective correct mode.

CV64/3D + Picasso96 + Warp3D stuff sort of works. CV64/3D + CGX4 + Warp3D does not, reason unknown. (at least in my test)

Fixed link , thank you Toni
https://download.abime.net/winuae/fi...nuae_4500b5.7z
https://download.abime.net/winuae/fi...ae64_4500b5.7z

[Rotareneg]

In the 4.5.0 betas (tested up to b5) 80 bit JIT FPU can cause FPU using programs to crash, tested with Sculpt4D, Real3D, and AIBB 6.5 FLOPS test. In 4.4.0 it works fine.

Config was A4000 68030 quickstart.

[Toni Wilen]

Virge 3D clipping fixed (Unexpected bug in original emulation code).

Fixes at least warptest corruption when texture is zoomed very close and needs vertical clipping.

Originally Posted by Rotareneg

In the 4.5.0 betas (tested up to b5) 80 bit JIT FPU can cause FPU using programs to crash, tested with Sculpt4D, Real3D, and AIBB 6.5 FLOPS test. In 4.4.0 it works fine.

Config was A4000 68030 quickstart.

I am going to take closer look to JIT changes later.

[Toni Wilen]

Originally Posted by Mad-Matt

Beta4 and Beta4fix keeps resulting in HPos hang/crash/error that jumps up in ever greater number the more you try to close the window. Really annoying to have to dig into task manager to quite uae.

Just a few seconds in a 3d prog seems to trigger it. No such issue encountered in Beta3.

Does https://download.abime.net/winuae/files/b/winuae.7z fix this? (Yesterday's update which fixed 3D clipping which could have corrupted memory)

btw, "3d prog" does not help much. P96 vs CGX makes a huge difference. Does warptest cause it too? (If not, "in a 3d prog" is even more useless )

[Mad-Matt]

Yeah the crash is still happening. The Engine demo that comes with Warp3D is enough to trigger it after a few seconds of moving about. This is with CGX v4 + Warp3D 4.2.

Whats interesting though is that the frame rate spikes to where it might be expected to be for a few short seconds before the crash.

[Toni Wilen]

Originally Posted by Mad-Matt

Yeah the crash is still happening. The Engine demo that comes with Warp3D is enough to trigger it after a few seconds of moving about. This is with CGX v4 + Warp3D 4.2.

Whats interesting though is that the frame rate spikes to where it might be expected to be for a few short seconds before the crash.

Warp3D + CGX simply crashes the Amiga here (memory corruption, jumping to non-existing memory or ROM address so most likely some dependency is wrong) so I can't really test it. Large Amiga programs/PC ports usually suck in error handling. Unfortunately.

[Toni Wilen]

winuae.7z updated. Run with log window open.

When/if it still crashes, does any "fifo full!!!!!" messages appear? ("fifo full.1." or "fifo full.2." are normal).