WinUAE 2.7.0 beta series

[Toni Wilen]

Originally Posted by PeterK

You're right Toni !!
Exactly, the missing 32 was the reason. I was using the 16 bit host mode. With 32 bit host mode the colors are correct now. Thanks a lot again.

Thanks for confirming. I guess I need to force 32-bit when "byteswapped" board is selected (any card except PicassoII/II+).

[mark_k]

Originally Posted by Toni Wilen

I can't see any other reason than output display mode being 16 bit (CL emulation only supports RGB byte swapping in 32-bit mode)

Check log and make sure it says "D3D9Ex: 640*480 main texture, depth 32", depth 32 is the important bit.

Ah. While my X desktop is 32-bit, for some reason the Fullscreen and 16/32 selections were initially blank. And the 16/32 selection drop-down box was initially empty! I had to select 1680x1050 in the Fullscreen dropdown before being able to select 32 for depth and that fixed the swapped colours. See pics. That may well be a Wine bug???

But I've also attached the config I've been using. That also has the bad expansion board diagnostic screen at boot time (bogus WinUAE graphics card I think).

Testing with Piccolo + EGS (without having installed Piccolo Freshup 1) there are some problems. Setting Workbench to 32 or more colours seems to work fine, but lower depths don't.

Depth 1: Workbench looks as it should but mouse pointer is not visible
Depth 2, 3, 4: Colours are wrong, looks monochrome.
Depth 4: As with depth 1-3 except there is some corruption at top of screen.
Depth 5-8: Seem to work fine.

The missing mouse pointer seems to only happen before you ever select a 32-or-more-colours EGS mode. For example, if you change WB to 32 colours then back to 4, the mouse pointer remains visible.

Next I set Workbench to EGS 640x480 32 colours, and opened Palette preferences (which appears on its own screen). On putting the Palette screen to the back, the Workbench appeared corrupted, see pic. Also this text in the log on each flip of the two screens: GFX SPECIAL BYTE PUT IO 8000 = 71

Finally, running EGS-NightShift (in EGS-Prefs drawer) the system hangs, log output CPU halted: reason = 3

[Toni Wilen]

Originally Posted by mark_k

Ah. While my X desktop is 32-bit, for some reason the Fullscreen and 16/32 selections were initially blank. And the 16/32 selection drop-down box was initially empty! I had to select 1680x1050 in the Fullscreen dropdown before being able to select 32 for depth and that fixed the swapped colours. See pics. That may well be a Wine bug???

They are blank if currently loaded config contains resolution/depth pair that does not exist in system. (Reason: if you save config, saved config still contains original resolution/depth values instead of defaulting to something)

Testing with Piccolo + EGS (without having installed Piccolo Freshup 1) there are some problems. Setting Workbench to 32 or more colours seems to work fine, but lower depths don't.

Known, again those annoying planar VGA modes with strange memory mapping setups.

[mark_k]

I noticed a problem with memory config in the latest winuae.zip (not present in 2700b1). With 2700b1 I could use my A2000-like config with 2M chip, 4M ZII fast, 16MB ZIII RAM. It seemed to work fine.

But with the latest winuae.zip, while having 8MB ZIII RAM works, 16MB doesn't. From the log output it looks like WinUAE tries to add a 16MB card at $200000 which doesn't work to well.

Fastcard: mapped @$200000: 16MB fast memory
   Card 1 (ZorroII) done.
ExceptionFilter Trap: 38701001 38700FFE 00000000
  D0 00000000   D1 00000020   D2 00000020   D3 00000000 
  D4 00000000   D5 00000000   D6 FFFFFFFF   D7 00000000 
  A0 00E80048   A1 00000B2C   A2 00E80000   A3 00000C00 
  A4 00E80000   A5 00000758   A6 00000A74   A7 000003B4 
USP  00000000 ISP  000003B4 SFC  00000000 DFC  00000000 
CACR 00000009 VBR  00000000 CAAR 00000000 MSP  00000000 
T=00 S=1 M=0 X=0 N=0 Z=1 V=0 C=0 IMASK=7 STP=0
01400FFE 0000 0000                OR.B #$ff,D0
Access violation! (68KPC=01400FFE HOSTADDR=38701001)
Your Amiga program just did something terribly stupid 38701001 PC=00000000
...

Log output with 8MB ZIII RAM configured instead:

Fastcard: mapped @$200000: 8MB fast memory
   Card 1 (ZorroII) done.
Filesystem: mapped memory @$e90000.
   Card 2 (ZorroII) done.
Z3MEM (32bit): mapped @$10000000: 8 MB Zorro III fast memory 
   Card 3 remapped 40000000 -> 10000000
   Card 3 (ZorroIII) done.

[Toni Wilen]

Originally Posted by mark_k

I noticed a problem with memory config in the latest winuae.zip (not present in 2700b1). With 2700b1 I could use my A2000-like config with 2M chip, 4M ZII fast, 16MB ZIII RAM. It seemed to work fine.

Fixed. Copypaste error in source memory bank cleanup.

ADDED: Piccolo ZIII is now actual ZIII board... Piccolo SD64 ZIII now shows cursor (wrong autoconfig data, was detected as 2M board and mouse cursor graphics is always in last 4k of VRAM)

[Toni Wilen]

http://www.winuae.net/files/b/winuae_2700b2.zip

Beta 2:

Cirrus Logic emulation updates:

- Fixed Piccolo SD64 Z2 wrong address space size.
- Piccolo is Z2/Z3 board.
- Fixed Z3 board subsize.
- Improved QEMU CL emulation glue code, VRAM banking/offsets supported.
- Some modes had invisible hardware sprite (mouse cursor).
- Implemented Piccolo/SD64/Spectrum vblank interrupts. Picasso96 won't hang anymore.
- Reset didn't reset all necessary graphics board variables.
- Force 32-bit host display mode if byteswapped board is selected (Any board except PicassoII/II+) Cirrus Logic emulation code supports byteswapping only in 32-bit mode.
- Note: Most VGA planar modes (16 colors or less) are not working. Includes also mouse cursor.

Other updates:

- Rewritten memory bank handling, moved all memory variables (start address, mask, size, etc) to addrbank stucture. Standard RAM bank code duplication replaced with macros.
- Debugger memwatch points now also detect hits in aliased addresses and only bank(s) that have watched addresses are redirected to debugger.
- CDTV SCSI option works again.
- Some uaenet.device improvements.
- It was not possible to write to Amithlon-style partitions.
- 5:4 aspect ratio had disappeared.
- CIA-A TOD counts (if vsync timed) on line 5 (PAL) or line 6 (NTSC), not line 0. (This is not 100% accurate yet, horizontal position is still wrong, will be improved in later betas)
- Installer update also coming, will be built with WiX (msi installer).

[amilo3438]

Hi,

Thanks for new WinUAE 2700b2 version,

except I cant even run it as it badly crashes on my PC/laptop ?!

Dont know whats could be a reason as WinUAE 2700b1 was running without problem.

[Toni Wilen]

Originally Posted by amilo3438

Hi,

Thanks for new WinUAE 2700b2 version,

except I cant even run it as it badly crashes on my PC/laptop ?!

Dont know whats could be a reason as WinUAE 2700b1 was running without problem.

Problem is Windows XP.

http://www.winuae.net/files/b/winuae_2700b2b.zip probably works.

[mark_k]

Beta 2 crashes if you start emulation with 68000 CPU and some Zorro III RAM configured. Obviously that's not a valid Amiga config but beta 1 and 2.6.1 don't crash then.

Also, for RTG cards for which WinUAE only supports a single fixed memory size, you could disable the memory slider in the settings window (so it is ghosted instead of just un-moveable).

[amilo3438]

Originally Posted by Toni Wilen

Problem is Windows XP.

http://www.winuae.net/files/b/winuae_2700b2b.zip probably works.

Now is OK. Thanks again.

[Toni Wilen]

Originally Posted by mark_k

Beta 2 crashes if you start emulation with 68000 CPU and some Zorro III RAM configured. Obviously that's not a valid Amiga config but beta 1 and 2.6.1 don't crash then.

I can't duplicate. Attach config.

Also, for RTG cards for which WinUAE only supports a single fixed memory size, you could disable the memory slider in the settings window (so it is ghosted instead of just un-moveable).

Too early. Not until I decide what to do with boards that have multiple sizes. (If it is possible to support all sizes or not)

[mark_k]

Here's the config I was using. Load it, then set ZIII RAM to 8MB and start emulation. Doing that gets WinUAE to crash for me.

The config has a Piccolo ZII graphics card, an HDF and shared folder. If I load the config, remove the HDF and shared folder, set ZIII RAM to 8MB and click Start there is no immediate crash; Kickstart 3.1 shows the insert disk screen. On pressing F12 to open the settings window, ZIII RAM shows as 0MB. If I change it to 8MB and click Reset WinUAE crashes.

Edit/update: another thing I noticed... if you set the RTG card to EGS Spectrum ZII and save the config, on loading the config the RTG card shows as Piccolo SD64 ZII.

[Toni Wilen]

Originally Posted by mark_k

Here's the config I was using. Load it, then set ZIII RAM to 8MB and start emulation. Doing that gets WinUAE to crash for me.

It had nothing to do with Z3. Non-uaegfx RTG enabled and hard reset crashed. Fixed.

Edit/update: another thing I noticed... if you set the RTG card to EGS Spectrum ZII and save the config, on loading the config the RTG card shows as Piccolo SD64 ZII.

Fixed.

[PeterK]

Originally Posted by Toni Wilen

....I can't see any other reason than output display mode being 16 bit (CL emulation only supports RGB byte swapping in 32-bit mode)......
Check log and make sure it says "D3D9Ex: 640*480 main texture, depth 32", depth 32 is the important bit.

A stupid question concerning the 16 bit output display host mode:
What are the 24 bit Amiga screenmodes really doing when the host is in 16 bit mode?
Are they clipping = reducing the the supported bits from RGB 888 to RGB 565 or something similar? Are these "TrueColor" Amiga screenmodes just a joke on a 16 bit host screen?

[Toni Wilen]

Originally Posted by PeterK

A stupid question concerning the 16 bit output display host mode:
What are the 24 bit Amiga screenmodes really doing when the host is in 16 bit mode?
Are they clipping = reducing the the supported bits from RGB 888 to RGB 565 or something similar? Are these "TrueColor" Amiga screenmodes just a joke on a 16 bit host screen?

At least uaegfx simply drops color component bits to fit it in 16-bits. CL emulation most likely does the same.

[PeterK]

Thanks for clarifying this, Toni

[mark_k]

When a graphics card is used, would it be possible to add an option to keep the native chipset display window open? So you'd have two windows, one for the chipset display and another for the graphics card output.

I've started to play with the NetBSD 5.2 HDF I uploaded a while ago. NetBSD supports RTG cards; the initial boot prompt is on the native display, but the rest of the boot text and console are on the Cirrus output. After logging in typing startx seems to do something, but the X display is not on the RTG output. I suspect it's actually on the native chipset display, but can't tell because the RTG console is still displayed. (After waiting a while, you can blindly type halt and X exits.)

Also, EGS apparently supports multiple displays so it might be useful to have both native and RTG displays shown at the same time. (According to the GVP-M EGS Spectrum page you could even use EGS with two Spectrum cards and have a single large EGS display spanning two monitors. )

A couple of things I noticed booting the NetBSD 5.2 HDF...

With Picasso II and II+, the emulation window doesn't switch to the RTG display automatically; it just shows a blank grey native screen.

The cursor is not visible sometimes. It doesn't seem to consistently appear (or not) with any specific card, but more or less randomly each boot. Happens with Piccolo, SD64, Spectrum.

[Toni Wilen]

Originally Posted by mark_k

When a graphics card is used, would it be possible to add an option to keep the native chipset display window open? So you'd have two windows, one for the chipset display and another for the graphics card output.

Answer is still no, simultaneous display would require really major internal changes (for too little gain for normal use case)

Option to manually switch between chipset and RTG may be possible.

With Picasso II and II+, the emulation window doesn't switch to the RTG display automatically; it just shows a blank grey native screen.

Log should have line "GFX SPECIAL BYTE PUT IO xxxx = yy" when it should switch screens (or multiple lines). Attach both switching to RTG and switching to chipset log lines. (Same for other board types too if there are differences). Thanks.

[mark_k]

For Picasso II/II+ booting NetBSD 5.2:
After the initial prompt, there is this log output (which should I assume switch to Picasso display):
GFX SPECIAL BYTE PUT IO 9000 = 01
GFX SPECIAL BYTE PUT IO 8000 = FF

At that point the native display is blank. After the OS boots, blindly logging in and typing startx, the X display eventually appears (on the native chipset display). No more GFX SPECIAL lines are printed up to that point.

Typing halt in the Xterm window results in
GFX SPECIAL BYTE PUT IO 8000 = FF
and the native display is blank again.

If you want to try yourself there are links to the test HDFs in this thread.

With Piccolo ZII:
GFX SPECIAL BYTE PUT IO 8000 = 10
NTSC mode lace V=60.0721Hz H=15665.4048Hz (228x262+1) IDX=11 (NTSC) D=0 RTG=0/0
GFX SPECIAL BYTE PUT IO 8000 = 00
GFX SPECIAL BYTE PUT IO 8000 = 20
...
GFXBOARD ACTIVE=1

The system boots showing console output on the Piccolo display. On typing startx there are no more GFX SPECIAL lines in the log.
Once X starts (you can't see the native display so wait a while) type halt. That results in
GFX SPECIAL BYTE PUT IO 8000 = 20
and text is output to the Piccolo display again.

[Toni Wilen]

Quick winuae.zip update, added RTG/chipset toggle input event. Temporarily also hardwired to Page Down for easier testing. (Switch from RTG to chipset is always possible, from chipset to RTG only if RTG mode is active)

EDIT: It wasn't that simple. Some more work to do..