WinUAE 3.4.0 beta series

[Arnie]

Sorry for the OT post ,but I need to clarify something here.

I thought the SX32 had the hard drive and yet you have the FMV module under 'Built in CD/HD controllers'.
Choosing this option adds an 'accelerator' option for Harddrives (which isn't detected)

Surely the FMV module should be in the Accelerators menu and the SX32 should be in both the Ide controllers and accelerators menus?

On a side not adding SX32 accelerator memory isn't working.

[Toni Wilen]

Originally Posted by Arnie

I thought the SX32 had the hard drive and yet you have the FMV module under 'Built in CD/HD controllers'.

Because it originally only had mainboard built-in CD/HD controllers. I forgot to rename the category title. (FMV is not technically built-in but it is close enough)

Choosing this option adds an 'accelerator' option for Harddrives (which isn't detected)

Surely the FMV module should be in the Accelerators menu and the SX32 should be in both the Ide controllers and accelerators menus?

FMV is not accelerator. It does not have replacement CPU.

SX32 IDE is boring. It has usual Gayle IDE and uses KS ROM driver (there is nothing IDE specific in its boot ROM) = enable normal A600/A1200 IDE to emulate it. Exact same thing.

Accelerator is never under SCSI/IDE controllers. Accelerator = whole board, including possible IDE or SCSI controller. (You can't separate them!)

On a side not adding SX32 accelerator memory isn't working.

SX32 memory has never worked properly. I listed it in some change log. It has stupid memory test that assumes specific hardware behavior I am not interested in debugging without real board.

[Arnie]

O.K. Thanks for the explanation

I did try the A600/1200 Ide but the HD wasn't detected with the 3.1 Installer disk. I used UAE in the end.

How would you add more than 8mb of memory to emulate the sx32pro, Z3 memory?

[Toni Wilen]

Originally Posted by Arnie

O.K. Thanks for the explanation

I did try the A600/1200 Ide but the HD wasn't detected with the 3.1 Installer disk. I used UAE in the end.

There is "conflict" with expansion panel Gayle IDE and Advanced chipset IDE settings. I haven't yet solved it without breaking backwards compatibility.

Workaround: untick Advanced chipset compatible setting, only then IDE enable checkbox stays ticked.

How would you add more than 8mb of memory to emulate the sx32pro, Z3 memory?

SX32pro uses non-standard address space for memory but easiest and closest is to use "CPU slot memory"

[Arnie]

Thanks for that info Toni. It works a treat

[Dr.Venom]

Originally Posted by Toni Wilen

Beta 5:

- Portaudio driver converted to pull mode.

I've been testing the updated portaudio modes and I'm getting sound crackling and these messages when using the portaudio wasapi mode in combination with low latency vsync:

20-714 [2273 227-204]: pull overflow! 16384 16384 16384

Currently there is sort of a workaround in that entering and exiting the gui a few times will "fix" it for a longer time (sound runs flawless then). I've attached a log where there's soundcrackling on startup and it being "fixed" after entering and exiting the gui.

This is from a quickstart configuration with only the portaudio wasapi mode and low latency vsync selected. From what I've tested the "No/Double/Triple buffer" selection has no influence on this.

[Toni Wilen]

Originally Posted by Dr.Venom

I've been testing the updated portaudio modes and I'm getting sound crackling and these messages when using the portaudio wasapi mode in combination with low latency vsync:

20-714 [2273 227-204]: pull overflow! 16384 16384 16384

Currently there is sort of a workaround in that entering and exiting the gui a few times will "fix" it for a longer time (sound runs flawless then). I've attached a log where there's soundcrackling on startup and it being "fixed" after entering and exiting the gui.

This is from a quickstart configuration with only the portaudio wasapi mode and low latency vsync selected. From what I've tested the "No/Double/Triple buffer" selection has no influence on this.

Does buffer size have any effect? "Real" buffer probably is at least 2x smaller now (no need for extra "sync buffer" anymore)

[Toni Wilen]

Originally Posted by Retro-Nerd

You could please try the SPS disks of Aladdin, Toni? With the latest few betas (also tried the one from yesterday) the game runs in app. 40fps (ingame!!) with Low Latency Vsync enabled. Other AGA games i've tested are fine so far.

Does b6 work any better? (different optimization setting)

[Dr.Venom]

Originally Posted by Toni Wilen

Does buffer size have any effect? "Real" buffer probably is at least 2x smaller now (no need for extra "sync buffer" anymore)

The overflows/crackling with Portaudio:Wasapi occur with all buffer settings (1-10). The only difference seems to be that the smaller the buffer size the more buffer overflows (the log gets filled very quickly with them at smaller settings) and the more audio crackling. The log I attached was with sound setting 8.

Note that the plain Wasapi mode has none of these issues on my setup, it seems purely related to the portaudio:wasapi and portaudio:wdm-ks modes.

Is there any (obvious) reason why entering and exiting the gui temporarily fixes the crackling/overflows?

[Retro-Nerd]

Originally Posted by Toni Wilen

Does b6 work any better? (different optimization setting)

Not with the b6 you've posted a few days ago. But with the winuae-beta.zip from today i get stable 50fps with the WASAPI driver. With DirectSound it's still the same.

[Toni Wilen]

Originally Posted by Dr.Venom

The overflows/crackling with Portaudio:Wasapi occur with all buffer settings (1-10). The only difference seems to be that the smaller the buffer size the more buffer overflows (the log gets filled very quickly with them at smaller settings) and the more audio crackling. The log I attached was with sound setting 8.

Note that the plain Wasapi mode has none of these issues on my setup, it seems purely related to the portaudio:wasapi and portaudio:wdm-ks modes.

Is there any (obvious) reason why entering and exiting the gui temporarily fixes the crackling/overflows?

Does it work now? (winuae.7z) PA mode used a bit too tight buffering and timing (same as wasapi ex)

[Dr.Venom]

Originally Posted by Toni Wilen

Does it work now? (winuae.7z) PA mode used a bit too tight buffering and timing (same as wasapi ex)

Works very well now Thanks!

[michaljarz]

Very thanks for JIT direct! Now it works (with 68K configs).
Also Wazp3D now working since beta 5 with WinXP

I tested other configs - with PPC Cyberstorm.
In 3.3.0 config Cyberstorm+JIT direct work (attachment 1), but in 3.4.0 (tested with latest build) not.
If I set "indirect", system started and when I change to "direct" then automatically returns to "indirect" (JIT: Reverting to "indirect" access, because canbang is zero!) (attachment 2).
If I set "direct" and run emulator then go to HALT7 (attachment 3).
If I set "direct" and switch off "UAE Zorro III" graphic card then system started, but returns to "Indirect" (attachment 4).

PS. I wanted out of curiosity to ask, which means the latest extension in the list of hardware, appearing when turns on "UAE Zorro III" graphic card?

[Toni Wilen]

Originally Posted by michaljarz

Very thanks for JIT direct! Now it works (with 68K configs).
Also Wazp3D now working since beta 5 with WinXP

I tested other configs - with PPC Cyberstorm.
In 3.3.0 config Cyberstorm+JIT direct work (attachment 1), but in 3.4.0 (tested with latest build) not.
If I set "indirect", system started and when I change to "direct" then automatically returns to "indirect" (JIT: Reverting to "indirect" access, because canbang is zero!) (attachment 2).
If I set "direct" and run emulator then go to HALT7 (attachment 3).
If I set "direct" and switch off "UAE Zorro III" graphic card then system started, but returns to "Indirect" (attachment 4).

Any accelerator that has maprom that needs address space mirroring (mainly blizzards, cyberstorms) = can't be JIT direct compatible. JIT does not support it.

JIT direct switched off: CPUBoard MAPROM
Switching JIT direct off!

PS. I wanted out of curiosity to ask, which means the latest extension in the list of hardware, appearing when turns on "UAE Zorro III" graphic card?

It is not an expansion. It shows last used Z3 address.

[emufan]

just posted in Addon-Card thread:
winuae beta 6 (2016.09.15) - in window mode, the floppy buttons are wrong now:
Power | HD | CD | N | 00 | 00
N is now df0: first "00" is "df1"

in Quickstart A500 it lists like this:
Power | HD | CD | | 00 | 00
empty space after "CD" is df0: hotspot

if already reported - just ignore.

[Toni Wilen]

Originally Posted by emufan

just posted in Addon-Card thread:
winuae beta 6 (2016.09.15) - in window mode, the floppy buttons are wrong now:
Power | HD | CD | N | 00 | 00
N is now df0: first "00" is "df1"

in Quickstart A500 it lists like this:
Power | HD | CD | | 00 | 00
empty space after "CD" is df0: hotspot

if already reported - just ignore.

I can't duplicate. Screenshot please.

[emufan]

Originally Posted by Toni Wilen

I can't duplicate. Screenshot please.

first the a500 quickstart, next my uber-pimped config, which didnt show this feature in final 3.3.0. red arrows point
to the hotspot for df0 "load an image" - 3rd is 3.3.0 with the same config as #2

[michaljarz]

N,DF0,DF1,... fields

[AMIGASYSTEM]

With 3.4.0 Floppy HD bootable by "Error Validating" during data loading, the same with floppy 3.3.0 no error.

[Toni Wilen]

http://www.winuae.net/files/b/winuae_3400b7.7z
http://www.winuae.net/files/b/winuae64_3400b7.7z

Beta 7:

- CD32 FMV ROM image was not loaded correctly from pre-3.4 config files.
- CD32 first CD audio track play attempt from the beginning was confused with TOC subchannel read.
- Z2/Z3 autoconfig data first byte's upper 5 bits was not editable. Only low 3 size bits should be forced (board size).
- Implemented non-autoconfig RAM board definition support. Added DKB Insider I/II.
- CD emulation didn't automatically load TOC when accessing the CD for the first time and program didn't request TOC previously. (Fixes Buddha ATAPI device not reading CD in non-image mode)
- Added manual (non-autoconfig) GUI memory board support. Any Z2 or Z3 board can be configured as manually mapped.
- Pre-KS1.2 boot support works again. Now expansion tree is build before emulation starts but pre-KS1.2 boot support obviously needs to know KS version before emulation starts but KS won't be loaded until emulation starts..
- Fixed harmless "SCSI command xx, no direction specified!" messages when attempting to play audio CD using hardware SCSI/ATAPI emulation.
- Fixed unstable Portaudio audio in vsync modes.
- Tecmar T-Card RAM works again. First Z2 RAM is emulated as Tecmar autoconfig RAM board.
- Shortcuts with SHIFT work again in optimized builds. It was buffer overflow (reading past end of buffer) very deep in input handling.
- Both SCRAM 500/2000 SCSI board variants emulated (DP8490V = 53C80 and NCR53C94).
- More ESP/NCR5394/FAS216 updates, FIFO in PIO mode only can contain max 1 byte. (SCRAM NCR53C94 fix)
- Otronic Filecard 2000/OSSI 500 emulation. (Official ROM is MIA, Aminet replacement rom works but it needs to be merged first)
- Automatically loaded default.uae with unplugged input device: set it to none, not layout A.
- Include CD/FMV audio caused infinite loop when playing CD audio.
- Disable ROM panel UAE MapROM option if accelerator board emulation is enabled. They are incompatible.
- Accelerator boards that required ROM but nothing was selected: crashed during expansion tree building.
- DD-only drive + turbo floppy mode + standard ADF HD floppy image: valid MFM was returned.
- Ignore all write attempts if standard ADF HD image in DD-only flpppy drive.

(some reported GUI related issues are not yet fixed)