Test report,
Realtek ALC883 @ nForce 430a (MCP61P)
MS-DOS 6.22
SBEMU ver.1.0b2

Output:

1Found sound card: Intel HDA
2Sound card IRQ conflict, abort.

Baron von Riedesel wrote on 2023-03-24, 13:01:

RayeR wrote on 2023-03-24, 01:29:

I tried SBEMU 1.0 beta 2b with Wolfenstein 3D. It started nicely with music but as I first shot the SFX sound scattered the screen by some unreadable error, speaker shortly beep and system hanged

I remember that kind of error, but thought it has been fixed long ago. Perhaps it's a different cause ...

Try the attached sbemu variant ( it's not compatible with the standard sbemu, so do not mix them ) and tell if the error occurs there as well.

I wish I could test that specific to see if it would help me with doom, but respvi.exe blanks my screen out (keyboard is responsive though) and my monitor loses its signal.
REM that line in start.bat, and it inits, but "No soundcard found", while in the two downloads on the first page, I get intel hda found.

Okay, I didn't have much time in these weeks, but I kept testing the program even though I didn't give a feedback.

Now, on my computer (Dell Inspiron 15, Intel HDA: Intel PCH (Panther Point) (80861E20) -> Realtek (10EC0282) (max 19
2kHz/24bit/8ch)), I didn't see any improvement since 3 or 4 test versions in context of these issues (although, I'm glad and it's impressive that at the meantime you've addressed other issues successfully):

Sound effects (AdLib and SB respectively) doesn't work with some games.

Shanghai 2: SB SFX tries to play once, produces only distorted noise and never tries to play again, ever. Every time a sound effect should play one voice of the music goes silent until the loop starts again. With AdLib, only the muting-of-a-channel issue occures, but still no SE.

Wacky Wheels: The setup finds the sound card, plays music, but no sound effects. If I try to adjust their volume, the system hangs. If I execute the game, the system hangs.

WWF Wrestlemania: Similar to Wacky Wheels.

Interestingly, Japheth's version performs better with Wacky Wheels, because it doesn't hang the system. One can play the game with music using Japheth's version, although, without sound effects.

Maybe some of his changes would worth being incorporated into the main project, becuase of its stability, also, an unloader would be very VERY useful for SBEMU, because once any other program touches the HDA, SBEMU should be reloaded, and while Japheth's simplified version can be unloaded an reloaded properly, yours can only be unloaded using active non-verbal violence, which is bad.

Interesting experiences:
1. SBEMU does things to the chipset that don't even reset by cold reboot, at least not on my computer. Once a game crushes badly, SBEMU seaces to work properly, will crush many other games, freezes, etc. until I boot up Windows7 and then reboot to DOS. Strange.
2. I could never get a no-Jemm, HDPMI32i-only setup work. Others reported it works for PM games. Not for me. Jemm is always required to work. Doesn't bother me, I just mentioned it.

For other users: Instead of JemmeX, Jemm386 is also usable. You have to compile the proper version of it, with the same version number Crazii's JemmeX has, it will have the same functionality. I compiled one for myself and posted it in the original thread of SBEMU, I have been testing SBEMU with both and they behave exactly the same way. I prefer Jemm386 because I primarily use an UMBPCI+XMGR setup and that CAN be extended with Jemm386 from the command line and Jemm386 can be theoretically unloaded once I'm done. In practice, this is currently only possibly with Japheth's version, because Crazii's original can only be unloaded via tricks and then Jemm386 doesn't feel like unloading anymore.

Would it be possible to include an optional IBM-speaker emulation in the future? Many computers with HDA don't have an actual PC-speaker, but a widget in the HDA in order to act as one. But if the HDA is "transformed" into a Sound Blaster, this won't work anymore and for games that use AdLib or SoundBlaster for music but beeper speaker for SE, SE no longer plays.

mikedebian wrote on 2023-03-24, 17:57:

REM that line in start.bat, and it inits, but "No soundcard found", while in the two downloads on the first page, I get intel hda found.

That's interesting - and there's a good chance to overcome this little hurdle. I changed the way how HDA devices are scanned - using PCI BIOS function "find device class" instead of "find device". The error message "no soundcard found" is misleading here, since most likely a HDA "card" was indeed found, but the MPXPLAY code refused to accept it because it is "digital only". In such cases one has to add cmdline option /DEV1 to make sbemu select the system's "second" HDA device.

... but respvi.exe blanks my screen out (keyboard is responsive though) and my monitor loses its signal.

That's even more interesting. These little tools just set/reset the PVI bit in CR4. In theory this should have no effect at all in v86-mode. So if the screen goes blank just by running SETPVI.EXE, it's a pretty weird system...
( btw, doom requires setpvi.exe, not respvi.exe ).

Baron von Riedesel wrote on 2023-03-24, 19:22:

That's interesting - and there's a good chance to overcome this little hurdle. I changed the way how HDA devices are scanned - u […]
Show full quote

mikedebian wrote on 2023-03-24, 17:57:

REM that line in start.bat, and it inits, but "No soundcard found", while in the two downloads on the first page, I get intel hda found.

That's interesting - and there's a good chance to overcome this little hurdle. I changed the way how HDA devices are scanned - using PCI BIOS function "find device class" instead of "find device". The error message "no soundcard found" is misleading here, since most likely a HDA "card" was indeed found, but the MPXPLAY code refused to accept it because it is "digital only". In such cases one has to add cmdline option /DEV1 to make sbemu select the system's "second" HDA device.

... but respvi.exe blanks my screen out (keyboard is responsive though) and my monitor loses its signal.

That's even more interesting. These little tools just set/reset the PVI bit in CR4. In theory this should have no effect at all in v86-mode. So if the screen goes blank just by running SETPVI.EXE, it's a pretty weird system...
( btw, doom requires setpvi.exe, not respvi.exe ).

It's an MSI X470 Gaming Plus Max (The name, I know...)
Powercolor Red Devil RX 6700XT (again, the name)

I suppose that as long as SETPVI.EXE makes me lose my video signal, and DOOM needs SETPVI.EXE, there's not much that can be done even though it would then detect my audio card.

Might it be to my config.sys/autoexec.bat?

https://pastebin.com/fsemSwiK

Here's a pastebin.

I REM the original autoexec.bat loading of sbemu, jload and hdpmi32i and used the ones in the new archive you posted. After it didn't work, I reverted my changes. You're looking at the reverted file.

Also, in config.sys I go in between jemmex and himem, because some times Jemmex will crap out. I tried with both.

mikedebian wrote on 2023-03-24, 19:44:

I suppose that as long as SETPVI.EXE makes me lose my video signal, and DOOM needs SETPVI.EXE, there's not much that can be done even though it would then detect my audio card.

You could try the MBF doom port - it doesn't need this PVI hack...

Kamerat wrote on 2023-03-23, 16:35:

The power state changed from 0x33 to 0x30 with the new version, but still no sound without running HDAplay first. […]
Show full quote

crazii wrote on 2023-03-23, 07:33:

This was discussed in another post in this thread, it is done, code not committed yet. This build should contain the fix Re: SB […]
Show full quote

Baron von Riedesel wrote on 2023-03-23, 03:30:

hdaplay indeed sets the power state, and probably sbemu should do as well: […]
Show full quote

hdaplay indeed sets the power state, and probably sbemu should do as well:

sc_inthd.c

Copy code to clipboard

1static int parse_output_path(struct intelhd_card_s *card,struct hda_gnode *node, int dac_idx)
2{
3	int i, err;
4	struct hda_gnode *child;
5
6	if(node->checked)
7		return 0;
8
9#if 1 /* set power state, seems necessary on (some?) notebooks */
10	snd_hda_codec_write(card, node->nid, 0, AC_VERB_SET_POWER_STATE, 0 );
11#endif
12	node->checked = 1;
13	if(node->type == AC_WID_AUD_OUT) {

Tried that on 1 laptop with HDA, it works.

This was discussed in another post in this thread, it is done, code not committed yet.
This build should contain the fix
Re: SBEMU: Sound Blaster emulation on AC97

The power state changed from 0x33 to 0x30 with the new version, but still no sound without running HDAplay first.

Copy code to clipboard

1Cold boot        0/ 16/0F05/0  - power state control=0x433
2SBEMU only old   0/ 16/0F05/0  - power state control=0x33
3SBEMU only new   0/ 16/0F05/0  - power state control=0x30
4HDAplay + SBEMU  0/ 16/0F05/0  - power state control=0x0

I updated another build, can you try this one: Re: SBEMU: Sound Blaster emulation on AC97

vorob wrote on 2023-03-12, 11:26:

Will check soon. For now I’ve checked another laptop, Compaq N410c with no native dos sound support. Worked flawlessly. Even made a video: https://youtube.com/shorts/rRuJxorpDVA

Thanks again!

Hi vorob and folks! Thanks for the work crazii!
I have the same laptop as vorob . All the stuff (jemmex, HDPMI32i, sbemu...) loaded without errors and detect ICH AC97, but games (like Duke3d) don't detect the emulated sound card.
Lauching Duke3d the log says something like"possibly conflict in DMA". I tested setsound.exe from Fatal Racing and got problems too.

Tested with MS-DOS 7.1 from Windows 98SE install and FreeDOS on a bootable pendrive.

Vorob, can you share your config.sys and autoexec.bat?
Did you make any specific adjustments to the settings?

This is my retro laptop for Windows 98 games (XP is also installed) and would be nice to have support for sound in DOS.

I just tested both SBEMU_VT8233Fix and Japhet's attached version.
Good new is that Wolf 3D SFX works now, both versions, but behaves differently in Dynablaster.

With old and SBEMU_VT8233Fix Dynablaster starts without sound, with Japhet's it hangs before setting GFX mode. And it hangs in really hard way - HDD LED remains lighting and even if HW reset button is pressed the BIOS cannot detect disks. Power off/on cycle is needed for recovery. I found this is caused by SETPVI.EXE. When I REM it in START.BAT then Dynablaster hangs too but a bit softer that HW reset is enough (no HDD LED lighting). But without SETPVI.EXE it doesn't work in Doom - hangs at timer init. This was probably expected. Wolf 3D works fine without SETPVI.EXE. I didn't know what PVI does, some short expl. here: https://www.oreilly.com/library/view/the-unab … 21lev1sec2.html Not sure why is needed trapping of CLI/STI executed by running program.

Some other quirks:
1) when I press CTRL+ALT+DEL with loaded SBEMU it hangs instead of reboot. I use FASTBOOT option for JEMM386 in config.sys

2) after SBEMU is loaded it delete BLASTER environment variable that I previously set in batch. I need to set BLASTER again after loading SBEMU. It was the last var. in env. list. Japhet's version don't do this.

Bruno128 wrote on 2023-03-24, 17:35:

Test report, Realtek ALC883 @ nForce 430a (MCP61P) MS-DOS 6.22 SBEMU ver.1.0b2 […]
Show full quote

Test report,
Realtek ALC883 @ nForce 430a (MCP61P)
MS-DOS 6.22
SBEMU ver.1.0b2

Output:

Copy code to clipboard

1Found sound card: Intel HDA
2Sound card IRQ conflict, abort.

You need disable some onboard devices to release IRQ, or try /i5 option, as noted in the first post. I will add more information to the error message in the next release.

RayeR wrote on 2023-03-25, 02:29:

I just tested both SBEMU_VT8233Fix and Japhet's attached version. Good new is that Wolf 3D SFX works now, both versions, but beh […]
Show full quote

I just tested both SBEMU_VT8233Fix and Japhet's attached version.
Good new is that Wolf 3D SFX works now, both versions, but behaves differently in Dynablaster.

With old and SBEMU_VT8233Fix Dynablaster starts without sound, with Japhet's it hangs before setting GFX mode. And it hangs in really hard way - HDD LED remains lighting and even if HW reset button is pressed the BIOS cannot detect disks. Power off/on cycle is needed for recovery. I found this is caused by SETPVI.EXE. When I REM it in START.BAT then Dynablaster hangs too but a bit softer that HW reset is enough (no HDD LED lighting). But without SETPVI.EXE it doesn't work in Doom - hangs at timer init. This was probably expected. Wolf 3D works fine without SETPVI.EXE. I didn't know what PVI does, some short expl. here: https://www.oreilly.com/library/view/the-unab … 21lev1sec2.html Not sure why is needed trapping of CLI/STI executed by running program.

Some other quirks:
1) when I press CTRL+ALT+DEL with loaded SBEMU it hangs instead of reboot. I use FASTBOOT option for JEMM386 in config.sys

2) after SBEMU is loaded it delete BLASTER environment variable that I previously set in batch. I need to set BLASTER again after loading SBEMU. It was the last var. in env. list. Japhet's version don't do this.

Yes pvi is another way to workaround doom freeze, which is discussed in the other thread linked in the first post.
SBEMU will set/override BLASTER env in case /I /A /D option is used, there's probably a bug in altering the BLASTER env.

What's the behavior of Ctrl alt del when SBEMU not loaded but HDPMI loaded ？ HDPMI will handle the key combination when a dpmi clientis active.

KormaX wrote on 2023-03-24, 18:13:

Maybe some of his changes would worth being incorporated into the main project, becuase of its stability, also, an unloader would be very VERY useful for SBEMU, because once any other program touches the HDA, SBEMU should be reloaded, and while Japheth's simplified version can be unloaded an reloaded properly, yours can only be unloaded using active non-verbal violence, which is bad.

Another way to handle this is to add reset option instead of unloading. Currently, most options(except /sc) can be changed after TSR. There's no reset yet but changes of /k (22050 to 44100 and vice versa) will actually do a reset.
EDIT: just checked the code again and /k won't cause a reset, I'll add another option to do that.
EDIT2: just added /R option to reset the soundcard driver. will be on next release.

KormaX wrote on 2023-03-24, 18:13:

Would it be possible to include an optional IBM-speaker emulation in the future? Many computers with HDA don't have an actual PC-speaker, but a widget in the HDA in order to act as one. But if the HDA is "transformed" into a Sound Blaster, this won't work anymore and for games that use AdLib or SoundBlaster for music but beeper speaker for SE, SE no longer plays.

Interesting suggestion but I'm not sure yet how/whether it could be done, probably another io trapping of speaker port.

Test report,
Realtek ALC655 @ ICH5
MS-DOS 6.22
SBEMU ver.1.0b2

Output:

1Found sound card: ICH AC97
2Real mode support: enabled.
3Protected mode support: enabled.
4OPL3 emulation enabled at port 388h.
5Sound Blaster Pro emulation enabled at Adress: 220, IRQ: 7, DMA: 1

Games tested:
Duke3D: sound and FM music are OK, reversed stereo.
Doom: sound and FM music are OK.
Disney’s Jungle Book: sound and FM music are OK.
Prince of Persia: freeze on “Loading…” screen.

Overall it is really good, sound is clear, fm is correct. Thank you!

Baron von Riedesel wrote on 2023-03-24, 20:34:

mikedebian wrote on 2023-03-24, 19:44:

I suppose that as long as SETPVI.EXE makes me lose my video signal, and DOOM needs SETPVI.EXE, there's not much that can be done even though it would then detect my audio card.

You could try the MBF doom port - it doesn't need this PVI hack...

It just randomly crashes at startup. Divide overflow errors.

I will follow the progress of sbemu and perhaps one day a work around will be issue 😀

Thanks for all the help though! Appreciated

MBH wrote on 2023-03-22, 20:18:

crazii wrote on 2023-03-22, 19:12:

MBH wrote on 2023-03-20, 15:41:

Thanks crazii, but it´s still not working. There is a progress, FM music is playing in Duke3d setup. It´s very disorted and way too slow, but at least a sign of live. For the soundeffects, there´s still the error with the wrong DMA channel. But I belive that´s a standard message if something is wrong and it is not for sure DMA related.

It's mostly possible that be interrupt that causing the problem. the only difference between mpxplay and sbemu is that sbemu enables interrupt. all the emulate SB's virtual IRQ & DMA is handled in the sound card interrupt. if interrupt is not working, games will report IRQ/DMA problem.
I tried another fix that I overlooked last time, it's updated here: Re: SBEMU: Sound Blaster emulation on AC97

I´m sorry, but still exactly the same problem. Nothing has changed. Duke3d setup complains about DMA and FM music ist disorted/too slow.

I have a Vt8233 in an HP T5520 thin client. I don't get any of the DMA conflict errors, but otherwise similar - MPXPlay works perfectly, but with SBEMU beta 2 no sound at all and with the VT8233FIX build, horribly slow and distorted sound for music and slightly less so for digital. This is for real mode (PlanetX3, Wolf3d) and protected mode (Descent). It reports as follows with /SCL (sorry for the shaky cam!).

Is there any new build I can try/test?

I should add beta 2 works absolutely perfectly on my daily driver (MSI B250i, HDA, i7-6600K)... just incredible!

crazii wrote on 2023-03-24, 22:15:

Kamerat wrote on 2023-03-23, 16:35:

The power state changed from 0x33 to 0x30 with the new version, but still no sound without running HDAplay first. […]
Show full quote

crazii wrote on 2023-03-23, 07:33:

This was discussed in another post in this thread, it is done, code not committed yet.
This build should contain the fix
Re: SBEMU: Sound Blaster emulation on AC97

The power state changed from 0x33 to 0x30 with the new version, but still no sound without running HDAplay first.

Copy code to clipboard

1Cold boot        0/ 16/0F05/0  - power state control=0x433
2SBEMU only old   0/ 16/0F05/0  - power state control=0x33
3SBEMU only new   0/ 16/0F05/0  - power state control=0x30
4HDAplay + SBEMU  0/ 16/0F05/0  - power state control=0x0

I updated another build, can you try this one: Re: SBEMU: Sound Blaster emulation on AC97

Thanks crazii. Unfortunately power state control is still at 0x30.

crazii wrote on 2023-03-25, 04:11:

Interesting suggestion but I'm not sure yet how/whether it could be done, probably another io trapping of speaker port.

You have to trap port 61h and PIT channel 2, so corresponding ports and behavior too. All trapped ports most likely should be passed to the real ports too. More info can be found here https://wiki.osdev.org/PC_Speaker

But so far Gravis Ultrasound would be much more interesting to be implemented -- its source code is already in dosbox which you are already using anyway.

crazii wrote on 2023-03-25, 03:17:

What's the behavior of Ctrl alt del when SBEMU not loaded but HDPMI loaded HDPMI will handle the key combination when a dpmi clientis active.

In this case (loaded only qpiemu and hdpmi32i) it works as expected - reboots from point of loading MBR (skipping the full POST) process. With sbemu it doesn't print any error just hangs and HW reset button have to be pressed.

But I found that loaded hdpmi prevents system shutdown via fdapm utility that should turn off PC via ACPI routine. It found DSDT, SSDT, GBTUACPI , port base as normally but the PC doesnt turn off within some minutes (normally turns off in 5-10s). But this has nothing to do with sbemu...

I didn't noticed the separate pvi thread so i will look at that to educate more about pvi.

RayeR wrote on 2023-03-25, 14:27:

I didn't noticed the separate pvi thread so i will look at that to educate more about pvi.

Since your laptop also crashes with PVI enabled, this hack will surely become a "last resort" only. Compared to that hack it's advantageous then to just trap CLI and replace it by a NOP - that would at least avoid any BIOS problems.

th3bar0n wrote on 2023-03-25, 12:19:

I have a Vt8233 in an HP T5520 thin client. I don't get any of the DMA conflict errors, but otherwise similar - MPXPlay works pe […]
Show full quote

MBH wrote on 2023-03-22, 20:18:

crazii wrote on 2023-03-22, 19:12:

It's mostly possible that be interrupt that causing the problem. the only difference between mpxplay and sbemu is that sbemu enables interrupt. all the emulate SB's virtual IRQ & DMA is handled in the sound card interrupt. if interrupt is not working, games will report IRQ/DMA problem.
I tried another fix that I overlooked last time, it's updated here: Re: SBEMU: Sound Blaster emulation on AC97

I´m sorry, but still exactly the same problem. Nothing has changed. Duke3d setup complains about DMA and FM music ist disorted/too slow.

I have a Vt8233 in an HP T5520 thin client. I don't get any of the DMA conflict errors, but otherwise similar - MPXPlay works perfectly, but with SBEMU beta 2 no sound at all and with the VT8233FIX build, horribly slow and distorted sound for music and slightly less so for digital. This is for real mode (PlanetX3, Wolf3d) and protected mode (Descent). It reports as follows with /SCL (sorry for the shaky cam!).

Is there any new build I can try/test?

I should add beta 2 works absolutely perfectly on my daily driver (MSI B250i, HDA, i7-6600K)... just incredible!

Yes, I just updated the zip : Re: SBEMU: Sound Blaster emulation on AC97
the problem is I don't have a hardware to debug, so currently only blind guess, hope it could be figured out this way.