1) My 2 cents to was fancinating story to read this thread. I read i as novel all 30 pages. I always made some summary checks in mind, it was like 1 year, after 1st video just 2 people on planet made it working and board where DOS PCI audio already was feasible.
I really doubt that much more people made it working outside this thread, i could be wrong.At 16 months LSS10999 made it working, but still compatibility is not great, but its nice to see more modern chipset (x99) to work.
Rasteri how many success stories were reported to you? How many people downloaded that PCB design?

2) Y PS2 on modern boards. I made it working one Gigabyte Z370 Gaming 2.0 board, other GB Z390 its not working and im not sure about other GB Z370 board, i requested it by that crazy GB request application and once got possitive answer that it would be added. There are is quite and lot undocummented bios changes which are actually happening with new BIOS update, lots of them is not gonna make it.
Again in theory if im not wrong PS/2 does no work as multithreaded device or asynchronously probably. If CPU just checking some IRQs list and reading signals.. It should not be hard make some small box, which would fastly as possible to switching PS/2 mouse and PS/2 keyboard mode, with some memory, to user not actually not feel that is turned on or off, its piece of cake in comparasion with this project. Im wrong about it?

3) Bios Editing DST, or how its named, it was long time ago, only hackingtosh thing, but there is Windows expert for it Dietmar, he made my board Z370 Gaming, WinXP + PAE friendly for example. After Win-raid all messages disappered (now its up again, include old messages) he is reachable on MSFN - https://msfn.org/board/profile/422112-dietmar/ , since Win-raid end..
He wrote 1280 messages there, so its very active, gifted, stubbord and friendly.
Im now trying to catch up on Vogons and read 1.5 years of news, i had to time to read his megathreads.

4)Idea: SB-EMU as frontend and real soucard as Backend.. If you are able to send send play sound on it,
without any fiddling with old chips. Im not expert in latency but maybe to can used Serial/Paralel/USB portor NIC and modified packet drivers, anything to feed it by data with good enough latency, to even spared Pata or Sata port. Simply anything which can send data to some PCB with ISA slot / or PCI slot to make it working for more modern soundcard on pci-E mbs.. Using these cheap PCI-E to PCI adapters, most common once have 2 PCI slots and some PCI-E to PCI already know chips no DMA,but otherwise out of box working adapters in Win9x+..
You can probably add some ASM sniffers(debugger) to see which commands is Windows 9x driver sending to cards make it working in Win9x when DirectSound is already inicialized.
Its same principle as Linux KVM semi virtualization, simply passthrough on DOS. Its probably way how Covox or OPL3LTP are already working with their much more limited old frontends(TSRs / per game patching), but it was before SBEMU era.
Advantage would be robust solution no soldering needed.

5) Original SB and SB Pro used IRQ7 as default, SB 16 later switched to IRQ5 as default. If you want to run all games you need both, especially old games without setup utilities require IRQ7. To make some old games playable and running with fast new shiny machines, there are other tricks neded. First test game withou Sound if it run at all, after is good idea make game working in same patch within Dosbox or 86Box and after port it to new machine. Also there is quite a lot game which has problem with Freedos, MS-Dos 6.22 or 7 is much more realible.
DR-DOS multitasking could be very nice for sound problems debugging once someone solved FAT32 problem (with FAT16 its already working-1 in "one windows", ASM level Debugger in "second window" without caring is debbuger is compatible with rendering(because its mosstly is not).. There version where is working multitasking fine and version where FAT32 is working fine.. but not both together and there is problem with MBR transition / or partion boot recornd, because MBR is needed to boot DR-DOS along side other OS (Rayer have article about it), but you want to use it on multiboot system with multiple partitions, geometry of disk is different and i dunno how to adjust it. Also quite a lot of game failuing just because too much freescape problem, MS-DOS is handling it, DR-DOS not, Dosbox has special -freescape or something like that parameter.

Details are here, i still hope that someone with some coding Dos skills will this it one day:
Dos multitasking / task switching tools research (MS 6.22/7.1/Freedos /DR-DOS), DRDOS compatibility (165+ games tested)

I created nice 700 MB Dos demo sound testing package its in Vogons library:
http://vogonsdrivers.com/getfile.php?fileid=1572
Which game needs which settings is there:
https://docs.google.com/spreadsheets/d/1VKadO … of=true&sd=true

Also there is list of DOS sound engines per game (see collumn D), once you have some game with same engine working, it there is very big chance to make it working with other game using same engine.

This could be also handy, list of games which are using VESA, where you can actually benefit from some faster CPU:
https://docs.google.com/spreadsheets/d/1qKTVV … dit?gid=0#gid=0

But be aware that after you fixed whole dos sound thing, all would be fine and easy. The right videocard for modern machine would be easy, its not, faster in modern OS, is often not faster in DOS, because all there not working power modes and booting at minimal clocks to safe some power..
https://docs.google.com/spreadsheets/d/1QPf4V … 0#gid=919349920

You are welcome to contribute though ask for editing rights through google account, you can add sheets note with you nick, its better than ugly sheets comments.

6) Lots of people care about placing such board in case and hold it as proper expansion card. There was made Big towers with 9/10 expansion slots, which can handle ATX /E-ATX board and 2/3 cards under. It took me Day to find them, its about size alone, lots of big cases have just 7 slots for ATX, now there are gone, these some era when people had quite a lot expension cards. I got one from coolermaster for my X58 machine with 3 GPUs, 2 sounds and some USB / Storage controller in plan. THey are quite unique and high end heavy duty. Now if chance to you can get them cheap, because their era endeded and people usually upgraded on something new. You can often get them full of outdated highend components and sell them by piece and keep the case. There duckduck them with magic keywords :"XL-ATX Case" to get some of them quickly, quite a high percententage of these cases have more than 7 full ATX slots.

Some examples: Lian Li - PC-P80 , Lian Li - PC-P80R, Thermaltake - Xaser VI, Thermaltake - Armor+ VH6000BWS, Mountain Mods - Ascension 7, HEC - 98 R9BB, Cooler Master - HAF X, SilverStone TJ11, Xigmantek Elysium, Rosewill Blackhawk Ultra. Once i wrote it here price will probably get a bit higher, but trying to be selfish..

7) Its possible to get new ISA slot somewhere or you need to get it from old MB?

😎 HWinfo is not bad, but its less stabe than Navratil SYstem info, lots of people ignoring it, but works often much better more modern than see was used too https://www.navsoft.cz/products.htm.

9) Im not expert, i cant solder or use osciloscope (i played with Quake modding in school instead actually learn how electorinc work, well except ASM, i some VESA engine when i was 13 or 14), but 1 thing seems very clear from which research, if some adress mess (mutiple devices on 1 adress), you have to use substractive decoding, there is not other way..without disabling conflicting devices.

ruthan wrote on 2024-12-08, 03:26:

1) My 2 cents to was fancinating story to read this thread. I read i as novel all 30 pages. I always made some summary checks in mind, it was like 1 year, after 1st video just 2 people on planet made it working and board where DOS PCI audio already was feasible.
I really doubt that much more people made it working outside this thread, i could be wrong.At 16 months LSS10999 made it working, but still compatibility is not great, but its nice to see more modern chipset (x99) to work.
Rasteri how many success stories were reported to you? How many people downloaded that PCB design?

Back then I had a hard time dealing with the adapter suddenly stop functioning. Tried and checked many things (especially about connection) and eventually the problem turned out to be the bridge chip I used being defective. Right now I have an adapter that's currently working stable for the past few months, powering a CT2950 with WavetablePi attached.

I mainly use it for DOS and WinXP there. Ironically, in a PAE-patched WinXP environment (up to 64GB), SB16 worked most stable compared to PCI/PCIe sound cards. My X-Fi Titanium PCIe would BSOD in several occasions involving MIDI (Soundfont), and in some occasions the driver would outright freeze the system during install leading to some filesystem corruption due to having to reboot the system in an unclean manner. ASUS cards are even more out of question as when I tried installing its driver on another PAE-patched WinXP system with a Xonar Essence ST, the system would always BSOD on driver install. Actually, the driver simply doesn't like PAE, as it also BSOD'd on the same system with Windows Server 2003 (on which PAE already worked without patching).

ruthan wrote on 2024-12-08, 03:26:

2) Y PS2 on modern boards. I made it working one Gigabyte Z370 Gaming 2.0 board, other GB Z390 its not working and im not sure about other GB Z370 board, i requested it by that crazy GB request application and once got possitive answer that it would be added. There are is quite and lot undocummented bios changes which are actually happening with new BIOS update, lots of them is not gonna make it.
Again in theory if im not wrong PS/2 does no work as multithreaded device or asynchronously probably. If CPU just checking some IRQs list and reading signals.. It should not be hard make some small box, which would fastly as possible to switching PS/2 mouse and PS/2 keyboard mode, with some memory, to user not actually not feel that is turned on or off, its piece of cake in comparasion with this project. Im wrong about it?

So far only ASRock offered the option to enable support for Y-Cables but not all PS/2 Y-Cables work as expected. The ones I'm currently using and working on such boards claimed to be for Advantech IPCs but they were not necessarily made by Advantech.

I wonder if it's somehow possible to add such support for other board vendors like ASUS and Gigabyte? Is this purely BIOS or it also needs specialized circuitry and SuperIO model?

ruthan wrote on 2024-12-08, 03:26:

3) Bios Editing DST, or how its named, it was long time ago, only hackingtosh thing, but there is Windows expert for it Dietmar, he made my board Z370 Gaming, WinXP + PAE friendly for example. After Win-raid all messages disappered (now its up again, include old messages) he is reachable on MSFN - https://msfn.org/board/profile/422112-dietmar/ , since Win-raid end..
He wrote 1280 messages there, so its very active, gifted, stubbord and friendly.
Im now trying to catch up on Vogons and read 1.5 years of news, i had to time to read his megathreads.

Honestly UEFI BIOSes are harder to tinker unlike older ones. So far I only succeeded in enabling the hidden "Advanced" menu and nothing else. Tried some other modifications but they all led to no boot stucking at certain POST codes.

While the hidden "Advanced" menu offers several options not originally accessible, the only thing I found relevant for DOS stuffs is a toggle regarding Gate A20.

ruthan wrote on 2024-12-08, 03:26:

4)Idea: SB-EMU as frontend and real soucard as Backend.. If you are able to send send play sound on it, without any fiddling wit […]
Show full quote

4)Idea: SB-EMU as frontend and real soucard as Backend.. If you are able to send send play sound on it,
without any fiddling with old chips. Im not expert in latency but maybe to can used Serial/Paralel/USB portor NIC and modified packet drivers, anything to feed it by data with good enough latency, to even spared Pata or Sata port. Simply anything which can send data to some PCB with ISA slot / or PCI slot to make it working for more modern soundcard on pci-E mbs.. Using these cheap PCI-E to PCI adapters, most common once have 2 PCI slots and some PCI-E to PCI already know chips no DMA,but otherwise out of box working adapters in Win9x+..
You can probably add some ASM sniffers(debugger) to see which commands is Windows 9x driver sending to cards make it working in Win9x when DirectSound is already inicialized.
Its same principle as Linux KVM semi virtualization, simply passthrough on DOS. Its probably way how Covox or OPL3LTP are already working with their much more limited old frontends(TSRs / per game patching), but it was before SBEMU era.
Advantage would be robust solution no soldering needed.

I don't have a good use case for SBEMU at the moment, as the systems on which it could be of some use are simply too modern (regarding onboard or discrete graphics' VGA capabilities).

But SBEMU already worked for some onboard and discrete sound cards as tested by some users.

ruthan wrote on 2024-12-08, 03:26:

5) Original SB and SB Pro used IRQ7 as default, SB 16 later switched to IRQ5 as default. If you want to run all games you need b […]
Show full quote

5) Original SB and SB Pro used IRQ7 as default, SB 16 later switched to IRQ5 as default. If you want to run all games you need both, especially old games without setup utilities require IRQ7. To make some old games playable and running with fast new shiny machines, there are other tricks neded. First test game withou Sound if it run at all, after is good idea make game working in same patch within Dosbox or 86Box and after port it to new machine. Also there is quite a lot game which has problem with Freedos, MS-Dos 6.22 or 7 is much more realible.
DR-DOS multitasking could be very nice for sound problems debugging once someone solved FAT32 problem (with FAT16 its already working-1 in "one windows", ASM level Debugger in "second window" without caring is debbuger is compatible with rendering(because its mosstly is not).. There version where is working multitasking fine and version where FAT32 is working fine.. but not both together and there is problem with MBR transition / or partion boot recornd, because MBR is needed to boot DR-DOS along side other OS (Rayer have article about it), but you want to use it on multiboot system with multiple partitions, geometry of disk is different and i dunno how to adjust it. Also quite a lot of game failuing just because too much freescape problem, MS-DOS is handling it, DR-DOS not, Dosbox has special -freescape or something like that parameter.

Details are here, i still hope that someone with some coding Dos skills will this it one day:
Dos multitasking / task switching tools research (MS 6.22/7.1/Freedos /DR-DOS), DRDOS compatibility (165+ games tested)

I created nice 700 MB Dos demo sound testing package its in Vogons library:
http://vogonsdrivers.com/getfile.php?fileid=1572
Which game needs which settings is there:
https://docs.google.com/spreadsheets/d/1VKadO … of=true&sd=true

Also there is list of DOS sound engines per game (see collumn D), once you have some game with same engine working, it there is very big chance to make it working with other game using same engine.

This could be also handy, list of games which are using VESA, where you can actually benefit from some faster CPU:
https://docs.google.com/spreadsheets/d/1qKTVV … dit?gid=0#gid=0

But be aware that after you fixed whole dos sound thing, all would be fine and easy. The right videocard for modern machine would be easy, its not, faster in modern OS, is often not faster in DOS, because all there not working power modes and booting at minimal clocks to safe some power..
https://docs.google.com/spreadsheets/d/1QPf4V … 0#gid=919349920

You are welcome to contribute though ask for editing rights through google account, you can add sheets note with you nick, its better than ugly sheets comments.

FYI I'm already using IRQ7 on my X99 setup, because it's the only IRQ I can use and that's after I disable onboard HDA which is on an exclusive PIRQ line that took it. I was simply unable to free IRQ5 on the board because it's wanted by many onboard PCI devices.

Your mileage may vary as modern boards don't offer the option to reserve IRQ and whether you can have easy access to IRQ5 or IRQ7 depends on the board design.

ruthan wrote on 2024-12-08, 03:26:

6) Lots of people care about placing such board in case and hold it as proper expansion card. There was made Big towers with 9/10 expansion slots, which can handle ATX /E-ATX board and 2/3 cards under. It took me Day to find them, its about size alone, lots of big cases have just 7 slots for ATX, now there are gone, these some era when people had quite a lot expension cards. I got one from coolermaster for my X58 machine with 3 GPUs, 2 sounds and some USB / Storage controller in plan. THey are quite unique and high end heavy duty. Now if chance to you can get them cheap, because their era endeded and people usually upgraded on something new. You can often get them full of outdated highend components and sell them by piece and keep the case. There duckduck them with magic keywords :"XL-ATX Case" to get some of them quickly, quite a high percententage of these cases have more than 7 full ATX slots.

Some examples: Lian Li - PC-P80 , Lian Li - PC-P80R, Thermaltake - Xaser VI, Thermaltake - Armor+ VH6000BWS, Mountain Mods - Ascension 7, HEC - 98 R9BB, Cooler Master - HAF X, SilverStone TJ11, Xigmantek Elysium, Rosewill Blackhawk Ultra. Once i wrote it here price will probably get a bit higher, but trying to be selfish..

For most standard ATX cases (7 expansion slots) dISAppointment can be installed just fine when paired with ITX or mATX boards.

The problem is ATX boards. One would have to use a high tower case with at least 10 expansion slots to accommodate.

ruthan wrote on 2024-12-08, 03:26:

7) Its possible to get new ISA slot somewhere or you need to get it from old MB?

Some suppliers may still have such slots in stock.

ruthan wrote on 2024-12-08, 03:26:

😎 HWinfo is not bad, but its less stabe than Navratil SYstem info, lots of people ignoring it, but works often much better more modern than see was used too https://www.navsoft.cz/products.htm.

On top of HWINFO I additionally used tools like PCICFG for detailed info about my system's PCI layout. There are some other tools like ASTRA.

ruthan wrote on 2024-12-08, 03:26:

9) Im not expert, i cant solder or use osciloscope (i played with Quake modding in school instead actually learn how electorinc work, well except ASM, i some VESA engine when i was 13 or 14), but 1 thing seems very clear from which research, if some adress mess (mutiple devices on 1 adress), you have to use substractive decoding, there is not other way..without disabling conflicting devices.

I don't think the problem is with subtractive decoding. On Intel chipsets (up to X99) you simply set up the address ranges you want to route to LPC (up to 4 can be set), and once you've routed the ranges used by the sound card to LPC it will work.

The current issue is AMD chipsets. The registers responsible for LPC address ranges are correct as in the documentations, but something is blocking access to almost all but a select few important addresses (like keyboard/mouse, POST card).

Couldn't hurt to email AMD and ask.
The worst they can do is sue you back to the stone age...

;p

Nah they'll probably help out

LSS10999 wrote on 2024-12-08, 10:42:

So far only ASRock offered the option to enable support for Y-Cables but not all PS/2 Y-Cables work as expected. The ones I'm currently using and working on such boards claimed to be for Advantech IPCs but they were not necessarily made by Advantech.
I wonder if it's somehow possible to add such support for other board vendors like ASUS and Gigabyte? Is this purely BIOS or it also needs specialized circuitry and SuperIO model?

I made dual PS/2 available on Gigabyte but not via Y-cable. It's not possible because the PCB designer was too lazy to route just 2 wires of second PS/2 port from superIO to connector (yes, near all superIO chips have 2 PS/2 ports). So I had to do their job myself by HW mod, but it works 😀
http://rayer.g6.cz/hardware/gap67ds3.htm#PS2

I give extra warning that GB/Asus often use superIO chips with non-standard pinout (compared to datasheet it usually have some numbering shift) - can be distinguished by extra model number/trailing chars. E.g. if you need to order replacement superIO chip you would need take special care of it's marking to buy the right one...

RayeR wrote on 2024-12-08, 16:18:

I made dual PS/2 available on Gigabyte but not via Y-cable. It's not possible because the PCB designer was too lazy to route just 2 wires of second PS/2 port from superIO to connector (yes, near all superIO chips have 2 PS/2 ports). So I had to do their job myself by HW mod, but it works 😀
http://rayer.g6.cz/hardware/gap67ds3.htm#PS2

Excellent bit of bodgery I must say!

BTW, often Y-splitters are wired backwards (green is keyboard, purple is mouse). There wasn't really a standard wiring, laptops usually had the mouse as the primary device and desktops have keyboard

I did a similar mod to an Asus board I have, single PS/2 that is either mouse or keyboard, with non-working Y splitter... all that was needed were two wires from SuperIO to the PS/2 port and now I have both mouse and keyboard. Simple case of not routing two traces as RayeR had.

Tiido, did your superIO have std or non-std pinout? I read there are special gigabyte and asus flavours and both have different pin numbering shift. I didn't mod asus board yet...

It matched exactly the datasheet I found, I don't remember what the part number is and I cannot check in any easy way either.

thanks for the responses, idk what i was thinking when i said i didn't know how to read the pin-outs.
i definitely can do that, but my confusion lied more between compatibility reasons, and you know what?
i checked out what Super I/O chip my mb has straight from the real thing and it's...

...an ITE IT8603E!

so i will check out if i can find any info online about it and then see if i can find an unpopulated resistor with the signal i'm missing on my mb

so i found the pin-out for my I/O chip... that was surprisingly quick and easy ngl...

where did i find it? well... i got it legally, that you can be sure about!

now all that's left is to read it and then figure out which direction is the chip actually facing on my mobo.

WOOOO PIN 67!!! *ahem* i managed to find the signal, thankfully.
now, i'll have to buy a multi-meter so i can find where it ends.
knowing how it went for you guys, i'm pretty sure it's on an unpopulated resistor, as it happens to be a trend with motherboard manufacturers apparently.

about the whole DIY assembly part...
i'm a little scared of messing it up considering my budget (or lack thereof), but at least i will try my best.

But with this you will find LDRQ used by superIO and not the second free LDRQ from PCH. Sure, as a last resort you can disconnect the LDRQ from super IO and use it for ISA bridge...

RayeR:
But with this you will find LDRQ used by superIO and not the second free LDRQ from PCH. Sure, as a last resort you can disconnect the LDRQ from super IO and use it for ISA bridge...

seems fine to me, i would do it anyways as long as i can make it work.
also i found this while thinking about ISA drivers for Windows 10:
(see attached pic) (it says "Standard PCI ISA Bridge" in spanish)

does this change anything? is it useful?
i hope so, that would be great

here's my list of all the signals i found for the project, just one last question...
which pin is PWRDN# supposed to be?

here's the asus pinout again, just in case

Heckett Heriot wrote on 2025-01-13, 13:52:

which pin is PWRDN# supposed to be?

PWRDN# is LPCPD#

I was expecting some comment to this idea:

4)Idea: SB-EMU as frontend and real soucard as Backend.. If you are able to send send play sound on it, without any fiddling wit […]
Show full quote

4)Idea: SB-EMU as frontend and real soucard as Backend.. If you are able to send send play sound on it,
without any fiddling with old chips. Im not expert in latency but maybe to can used Serial/Paralel/USB portor NIC and modified packet drivers, anything to feed it by data with good enough latency, to even spared Pata or Sata port. Simply anything which can send data to some PCB with ISA slot / or PCI slot to make it working for more modern soundcard on pci-E mbs.. Using these cheap PCI-E to PCI adapters, most common once have 2 PCI slots and some PCI-E to PCI already know chips no DMA,but otherwise out of box working adapters in Win9x+..
You can probably add some ASM sniffers(debugger) to see which commands is Windows 9x driver sending to cards make it working in Win9x when DirectSound is already inicialized.
Its same principle as Linux KVM semi virtualization, simply passthrough on DOS. Its probably way how Covox or OPL3LTP are already working with their much more limited old frontends(TSRs / per game patching), but it was before SBEMU era.
Advantage would be robust solution no soldering needed.

Its is no go and some compatible old / chipset cpu is needed to initialize / process sound through LPC, what about to actually add such subsystem directly to this addon? I mean some low power cheap Atom or any old x86 cpu.. and if not lowlevel ASM stuff is not enough to process play sound / music effect, in theory with faster cpu, there could be whole minimal modern Linux stack with sound, which can redirect to PC through SW, there could be even just audio jack out cable hardwire it into modern sound care Line in.. and its still too complicated whole module would be connected through serial / parallel / usb.. and it would work like USB soundcard, which is running even on WIndows 98, so not that far from Dos.

Tiido wrote on 2024-12-04, 21:13:

I finally ordered some PCBs and among them is this :
I'm not quite sure when I get to assemble it, but it shouldn't be too far in the future. I had my first exam today which went well and another will come on friday which should go well too and after that there's only few more weeks of school left ~

Tiido, did you finish your 3-slot PCB assembly and test it with multiple ISA cards plugged in? I'm just curious if there's no some problem with limited ISA bus ports driving capabilities of the bridge. Typical use case is just a single slot on an industrial MB. I didn't deeply study the datasheet if it can safely drive more cards in more slots...

RayeR wrote on 2025-01-15, 20:58:

Tiido, did you finish your 3-slot PCB assembly and test it with multiple ISA cards plugged in? I'm just curious if there's no some problem with limited ISA bus ports driving capabilities of the bridge. Typical use case is just a single slot on an industrial MB. I didn't deeply study the datasheet if it can safely drive more cards in more slots...

The official Fintek appnote has 3 slots.

(It claims to be able to source/sink 24mA which is about the same as a 74LS245 (as used in IBM PC). So I reckon it should work for a dozen or more cards)

Then in should be OK... except lack of enough IO windows in LPC bridge for multiple cards but could be reprogrammed on the fly...

RayeR wrote on 2025-01-15, 20:58:

Tiido, did you finish your 3-slot PCB assembly and test it with multiple ISA cards plugged in? I'm just curious if there's no some problem with limited ISA bus ports driving capabilities of the bridge. Typical use case is just a single slot on an industrial MB. I didn't deeply study the datasheet if it can safely drive more cards in more slots...

I have not finished it, I don't have slots to put on it 🤣
But I expect there to be no issues with several cards, even if they are ancient ones with bipolar and not CMOS inputs.