... Nope. Just tried plugging my Sound Blaster 16 (non-PnP) into the ISA slot and it was not picked up. DIAGNOSE could not detect I/O (220h), not even FM (388h), despite the bits responsible for allowing those ranges to be decoded were set.

I probably need to look at other documentations of this chipset, as I'm afraid that register programming guide won't be more helpful until I can find out the true nature of this "firewall". Figuring this out will probably also solve the issue with games not detecting Sound Blaster for PCI sound cards, even though their respective TSRs are loaded and resident.

Port 80h is obviously allowed by this "firewall" so debug cards can work. For PCI it works out-of-box, while for ISA (behind the Fintek) the bit needs to be manually enabled.

Another idea - Im just reading https://xairy.io/articles/thinkpad-xdci where author managed to enable hidden device mode USB interface. One of the things he tried was looking how bios disables/enables device he is interested in. Intel SKD UEFI SDK leaked and you can read bios code in source form without worrying about decompiling UEFI mess. Relevant LPC stuff might be in bios sdk.

Does anyone have datasheets/documentations for nVidia nForce chipsets? Unlike Intel and AMD, I searched everywhere and could not find anything useful.

Some later nForce boards (mainly for AMD CPUs, may also apply to those for Intel CPUs as well) have footprints for a TPM header but not actually soldered. Additionally, some parts such as resistors between PCICLK_TPM and the actual PCICLK pin on the chipset are also missing. Not sure how many ohms I should solder on the respective pads, from what I could find in some schematics of other boards the value seems to be between 10-22 ohms.

I'm currently experimenting with a MCP68-based (N68) board from ASUS that seems to expose LDRQ1# per its boardview, as well as having footprints for a TPM port.

With the help of "setpci" from Linux I'm able to dump all the register contents from the chipset's LPC bridge (bus 0 device 1 function 0), but without any documentation, I won't be able to easily figure out registers other than those common to every PCI device.

Unlike AMD chipsets, nForce chipsets do not have the "firewall" problem. Legacy I/O can reach PCI sound cards via TSR without issues, just no proper DMA over PCI.

That serial resistors value doesn't need to be precise I would use 22R or 33R, what you find...
Good to know that nforce passes IO properly. Maybe some chipset compatability list could be added in the 1st post or somewhere else where others could find it.

This might have been asked many times, but there is no way to buy these pre-assembled for those who cannot solder for whatever reason?

I bought me an ALIX 3D3 for extremely cheap which has an LPC output so I thought oh I could just buy that, get me a dISAppointment, and off ya go.
Its an AMD Geode LX800 with 256MB DDR1, 85fps on quake default, 186fps on doom. (using dosbench)

Furthermore, running the driver exe gives me "Got 98DF0, expected 8086.". So from the mention of intel LPC controllers not being found, only Intel is supported by default?

Edit: Typo's, and I have to note after working on this thing all day, the BIOS bugs meaning mouse support will never happen (rs232 gives no power, usb keeps getting reset if using a mouse) is a bit of a hit to this things possibillities...

So, just taking a casual glance at the manual for my Recording PC's motherboard, here's the pinout of the TPM header:

<snipped, since it was a LIE anyway!>
20 LDRQ#

So would I be right in thinking that I could drop an LPC2ISA board in this lovely Skylake (I could go Kaby, but that seems a bit lairy) box?

InTheStudy wrote on 2024-03-09, 20:07:

So, just taking a casual glance at the manual for my Recording PC's motherboard, here's the pinout of the TPM header: […]
Show full quote

So, just taking a casual glance at the manual for my Recording PC's motherboard, here's the pinout of the TPM header:

1 LCLK
2 GND
3 LFRAME#
4 <(KEY)>
5 LRESET#
6 +5V
7 LAD3
8 LAD2
9 +3.3V
10 LAD1
11 LAD0
12 GND
13 SMB_CLK
14 SMB_DAT
15 +3V Stdby
16 SERIRQ
17 GND
18 CLKRUN#
19 LPCPD#
20 LDRQ#

So would I be right in thinking that I could drop an LPC2ISA board in this lovely Skylake (I could go Kaby, but that seems a bit lairy) box?

I don't think so. Skylake and onwards don't have any form of LPC DMA anymore. The (possibly emulated from eSPI) LPC bus doesn't have LDRQ# and the datasheet does not have a single mention of 8237 controller.

Broadwell (9 series), the generation right before Skylake, is the last to feature LPC DMA and 8237.

While the 20-pin TPM header is mostly standardized (so a TPM from any vendor would work), the TPM itself doesn't use LDRQ#.

LSS10999 wrote on 2024-03-10, 02:13:

I don't think so. Skylake and onwards don't have any form of LPC DMA anymore. The (possibly emulated from eSPI) LPC bus doesn't have LDRQ# and the datasheet does not have a single mention of 8237 controller.

That's a surprise for a couple reasons. First, it's not like SuperMicro to be misleading in their manuals, and second - this is not the only vendor of a C236 motherboard which lists LDRQ on the TPM header. Plus, Dell mentions "two 8237 DMA controllers" in their specs of C236 based boards.

That said I've just found the C230 datasheet and you're completely right of course:

"LPC: 24 MHz, No DMA"

Blast you, SM and Dell! False hope! So, taking that to the logical conclusion - assuming SuperMicro isn't also lying about their TPM headers on boards which actually HAVE that pin on the PCH... Peak Intel board for ISA is the X10SRA? The dual socket board doesn't list DIRQ on the header, but it's on the X10SLH (single socket 1150, PC3 32GB) and on the X10SRA (single socket 2011, PC4 512GB).

And since those take up a full 7-slot board for their 42PCIe lanes, I guess ISA will have to sit in the 5.25" bay? It's annoying though, I have enough kit in the junkpile at work to set up one of those boards with 10 cores and 64GB ram; but the boards themselves are £400 used on eBay. The idea of getting an ADLIB card isn't worth that much to me, especially when a Retrowave USB-C OPL3 is $50.

MB manual writers obviously use ctrl+c/v too often 😀

DutchComputerKid wrote on 2024-03-09, 14:21:

This might have been asked many times, but there is no way to buy these pre-assembled for those who cannot solder for whatever r […]
Show full quote

This might have been asked many times, but there is no way to buy these pre-assembled for those who cannot solder for whatever reason?

I bought me an ALIX 3D3 for extremely cheap which has an LPC output so I thought oh I could just buy that, get me a dISAppointment, and off ya go.
Its an AMD Geode LX800 with 256MB DDR1, 85fps on quake default, 186fps on doom. (using dosbench)

Furthermore, running the driver exe gives me "Got 98DF0, expected 8086.". So from the mention of intel LPC controllers not being found, only Intel is supported by default?

Edit: Typo's, and I have to note after working on this thing all day, the BIOS bugs meaning mouse support will never happen (rs232 gives no power, usb keeps getting reset if using a mouse) is a bit of a hit to this things possibillities...

AMD Geode? Interesting. So there are boards with means to expose LPC directly, even predate TPM...

Does your board also expose a LDRQ# signal? If so then you should be able to connect the adapter to the bus with little effort.

The software part is specific to Intel. The host LPC bridges of different chipset manufacturers are placed and configured differently so one would have to write different code to handle them.

You need to inspect your PCI topology and find where your chipset's LPC (ISA) bridge is. You'll also need the documentation of your chipset to know more about the configuration registers of the bridge so you can tweak it to enable the options required for sound cards to work.

Btw AMD Geode may have VSA sound emulation built-in so SB music could work out of the box via onboard sound codec if present. I have one thin client with such feature and sfx works well there but the thin client completly lakcs PS/2 for KB+mouse - have only USB ports without legacy support so it's not much usefull for DOS anyway...

[/quote]

AMD Geode? Interesting. So there are boards with means to expose LPC directly, even predate TPM...

Does your board also expose a LDRQ# signal? If so then you should be able to connect the adapter to the bus with little effort.

The software part is specific to Intel. The host LPC bridges of different chipset manufacturers are placed and configured differently so one would have to write different code to handle them.

You need to inspect your PCI topology and find where your chipset's LPC (ISA) bridge is. You'll also need the documentation of your chipset to know more about the configuration registers of the bridge so you can tweak it to enable the options required for sound cards to work.
[/quote]

To all of the above, yes the ALIX 3D3 has all the pins for the dISAppointment exposed and supposedly working, since PCengines shared the shematics. (neat!)
I dont know why they chose LPC for the ISA bus either. And for software, man that makes bigger industrial systems (Like my 13 slot IBM industrial computer, AMD K6 PICMG) sound like a better idea but its so damn big and heavy. Not to mention, to build a smaller one with ISA comes down to 200 euro or more. Whereas the alix was 15, since the seller had no idea what it was/is.

RayeR wrote on 2024-03-12, 13:10:

Btw AMD Geode may have VSA sound emulation built-in so SB music could work out of the box via onboard sound codec if present. I have one thin client with such feature and sfx works well there but the thin client completly lakcs PS/2 for KB+mouse - have only USB ports without legacy support so it's not much usefull for DOS anyway...

As for this, yes it should, kinda. It does have AC97, but externally with a Realtek ALC203, which SBEMU does not support and the BIOS does not seem to have any SB settings. It would make things easier, but its no option.

So if the entire driver sitution would need a rewrite, nevermind manufacturing costs, does not seem that effective of a solution anymore. It was going to use an Aztech Sound Galaxy Nova 16 Extra II.

LSS10999 wrote on 2024-03-12, 09:41:

AMD Geode? Interesting. So there are boards with means to expose LPC directly, even predate TPM... […]
Show full quote

AMD Geode? Interesting. So there are boards with means to expose LPC directly, even predate TPM...

Does your board also expose a LDRQ# signal? If so then you should be able to connect the adapter to the bus with little effort.

The software part is specific to Intel. The host LPC bridges of different chipset manufacturers are placed and configured differently so one would have to write different code to handle them.

You need to inspect your PCI topology and find where your chipset's LPC (ISA) bridge is. You'll also need the documentation of your chipset to know more about the configuration registers of the bridge so you can tweak it to enable the options required for sound cards to work.

To all of the above, yes the ALIX 3D3 has all the pins for the dISAppointment exposed and supposedly working, since PCengines shared the shematics. (neat!)
I dont know why they chose LPC for the ISA bus either. And for software, man that makes bigger industrial systems (Like my 13 slot IBM industrial computer, AMD K6 PICMG) sound like a better idea but its so damn big and heavy. Not to mention, to build a smaller one with ISA comes down to 200 euro or more. Whereas the alix was 15, since the seller had no idea what it was/is.

RayeR wrote on 2024-03-12, 13:10:

Btw AMD Geode may have VSA sound emulation built-in so SB music could work out of the box via onboard sound codec if present. I have one thin client with such feature and sfx works well there but the thin client completly lakcs PS/2 for KB+mouse - have only USB ports without legacy support so it's not much usefull for DOS anyway...

As for this, yes it should, kinda. It does have AC97, but externally with a Realtek ALC203, which SBEMU does not support and the BIOS does not seem to have any SB settings. It would make things easier, but its no option.

So if the entire driver sitution would need a rewrite, nevermind manufacturing costs, does not seem that effective of a solution anymore. It was going to use an Aztech Sound Galaxy Nova 16 Extra II.
[/quote]

If I remember well it can be inspected inside BIOS image if it contains VSA or not according to a specific header. In my case I don't need any driver or config utility, no SBEMU. I just setup BLASTER env (typical IO 220h, IRQ5) variable and it plays... Not sure how many Geode platforms was equipped with VSA. It was some SM mode binary blob provided directly by AMD to BIOS writers that may or may not include it.

I have an alix 2d2 I keep meaning to test, unfortunately it's the one without a graphics card so I first have to build some miniPCI to PCI adapters (another stalled project...)

I made miniPCIe to PCIe x1 adapter for tinyLlama, not tested yet... Hm, miniPCI woul be a much more wires but less speed... coul be doable on a 4L PCB..

I just did some further tests on the supposedly working dISAppointment and found out it might not be 100% usable. This might have also rendered several questions I posted before invalid...

This is on an ASRock X99M Killer 3.1. I connected all but LDRQ# to dISAppointment and on the surface it seems to work. My programs can correctly detect the Fintek bridge (and all its registers), and the POST card I used for testing the ISA slot is working as I'm getting port 80 output there.

However, ISA PnP sound cards can't be detected even after I enabled the necessary ranges. Tested several known working cards I have and none were detected. I'm not sure what might be causing this. I think most of the LPC/ISA signals are already working considering the fact that the POST card already worked.

I haven't connected LDRQ# yet, as I need to remove the heatsink and solder a wire underneath it to bring out the unused LDRQ1#. However, I don't think a working ISA DMA is required for ISA PnP detection.

By "can't be detected" you mean a soundacd configuration utility from card manuf. or unisound or some sb testing utility or game? In my case the ess1869 was not detected and configured by esscfg until I enabled 6xx IO range. Without config it didn't play.

RayeR wrote on 2024-03-22, 01:50:

By "can't be detected" you mean a soundacd configuration utility from card manuf. or unisound or some sb testing utility or game? In my case the ess1869 was not detected and configured by esscfg until I enabled 6xx IO range. Without config it didn't play.

I'm using UNISOUND. Tried a CT2950 and a YMF718-based card. Both worked on RUBY-9719VG2AR, but not here...

rasteri wrote on 2024-03-15, 20:01:

I have an alix 2d2 I keep meaning to test, unfortunately it's the one without a graphics card so I first have to build some miniPCI to PCI adapters (another stalled project...)

Ah a D2D, so with tinyBIOS, you likely wont get any VGA support at all. You have to use one with Award for that. TinyBIOS throws everything over the serial port.
Propreitary PCEngines firmware, so it likely wont work at all.