PCie-to PCI, PCI to ISA, # of slots multipliers - bridges, risers, backplanes, research, especially for DOS, WIP.

Discussion about old PC hardware.

Re: PCie-to PCI, PCI to ISA, # of slots multipliers - bridges, risers, backplanes, research, especially for DOS, WIP.

Postby Doomn00b » 2019-5-08 @ 16:53

LSS10999 wrote:
Doomn00b wrote:I have a question here... I have a Asus M3A78-CM rev 1.01G motherboard in my possession, it's an AMD board, it features an SB700 southbridge chip - just to make sure, I had a look at the specs, and unless I've misunderstood it, this Southbridge has very complex DMA support, with Midi and Adlib mentioned:

http://developer.amd.com/wordpress/medi ... b_1.00.pdf

Is it me, or can this MB use PCI soundcards which will work quite well in pure dos??

Here's the specs for the entire board:
http://static.highspeedbackbone.net/pdf ... Manual.pdf

AMD rarely gets mentioned in these threads, so perhaps it's a good idea to take a closer look at their chipsets. A common thing I keep hearing, is that AMD has terrible support for ISA and DOS PCI sound, but as Ruthan says... "there are many urban legends spouted".


EDIT:

Some of the urban legends are indeed true... the SB700 line of southbridge chips are all good, but later chipsets have various types of DEFECTIVE LPC controllers - so, SB750S is the last one to definitively have a functional and rich LPC - complete with advanced Super-I/O.


Modern AMD chipsets are a complete no-go for PCI audio cards. I once had boards of each generation (700/800/900) and none worked, not to mention on those chipsets, the games still can't detect sound blaster even after I loaded the TSR (and confirmed the TSR is indeed loaded).

EDIT: I briefed the datasheet a bit and did find references of legacy audio. It seems starting from pages 171 (SMBus Module and ACPI Block) there were references about enabling some legacy audio hardware I/O addresses (Adlib 388-389, MIDI 300-330, Sound Blaster 220-280, WSS 530/604/E80/F40) to trigger SMI# (System Management Mode).

This makes it possible on SB7xx to emulate legacy audio using system management mode hacks, but it doesn't provide any means to make existing hardware's legacy TSRs usable (as they don't and can't use SMI). One needs to develop specific SMM code against the chipset and the target sound card then integrate it to the BIOS to get legacy audio working. For this, it might be more practical to target the onboard audio than targeting a specific discrete audio card (as if targeting a discrete card you'll probably need to modify the BIOS code every time you change stuffs, unless if it's possible to put soundcard-specific code in a higher ring from SMM(ring -2)). I'm not an expert in SMM so I'm not entirely sure). Again, this is only a possibility.

EDIT 2: Did not find any similar references on SB8xx register reference guide (and there's no SB9xx databook out yet), but on SB8xx databook it has references about LPC (which isn't useful on its own, SB7xx also has such mentions). The audio-related SMI# stuffs are so far only mentioned on SB7xx chipset documents.

PS: It has been possible to use SMI# to emulate legacy audio and other stuffs before (see Virtual Subsystem Architecture (VSA)).


EXCELLENT find, my dude!

So, there might be a form of virtualization available for legacy devices, eh? We just have to figure out more about it, lol! Custom Bios's are a thing, so once we know more, we shalt not let that stop us.

In the meantime, I've been checking out various AMD 790FX and 790GX bords, and a few look quite promising! One actually has 7 slots, and would hence be suitable - I believe it takes at least one AM3+ cpu as well...


MSI 790FX-GD70

DFI DK 790FXB-M3H5
User avatar
Doomn00b
Newbie
 
Posts: 39
Joined: 2018-9-14 @ 14:02
Location: The Freezing north

Re: PCie-to PCI, PCI to ISA, # of slots multipliers - bridges, risers, backplanes, research, especially for DOS, WIP.

Postby ruthan » 2019-5-08 @ 18:19

These dont seems to be big deal, there are same age or older than X58/X79.. which are working out of box.
Phenom II wasnt great cpu. Maybe there is advantage, that you can get them cheap..
Im old goal oriented goatman, i care about facts and freedom, not about egos+prejudices. Hoarding=sickness. If you want respect, gain it by your behavior. I hate stupid SW limits, SW=virtual world, everything should be possible if you have enough HW.
User avatar
ruthan
Oldbie
 
Posts: 1016
Joined: 2013-3-07 @ 04:01
Location: Schwarz Wald-from France to Ukraine, from Denmark to Austria. Celts+German+Slavs melting pot.

Re: PCie-to PCI, PCI to ISA, # of slots multipliers - bridges, risers, backplanes, research, especially for DOS, WIP.

Postby cyclone3d » 2019-5-08 @ 18:49

ruthan wrote:I looke at that Magma cases:
https://www.onestopsystems.com/product/ ... -expansion

But even if they would work they are expensive as hell, simply company stuff where you need it and prices are not matter.. and we will not know if they are working for sound cards.. but contruction seems pretty simple some special card and case with backplate own PSU..

Regardless i tried to write them, but no lock:
MS DOS is not supported and not compatible with 3slot PCI Expansion unit,

Thanks,
OSS Technical Support


I've been getting my units off of eBay. I wouldn't dream of spending the amount they charge for new units.

As far as the DOS compatibility goes, we shall see. I'm guessing that is just their standard response since I'm guessing that they really have 0 incentive to test with DOS.
These setups are also driverless so I don't see why they wouldn't work with DOS. They should just be transparent to the OS and just be seen as a PCI bus to the host system.

I did try the PC-card with the 1-slot on one of my Toshiba notebooks but it didn't like it. It had some sort of resource conflict that was messing with the video card. The PC-card I have is also the earlier version which apparently has issues with some cards that these setups are normally used for. I'll try it again with some different laptops to see what the results are and post back.

I also have one of the PCI host cards on the way so I can test with a regular desktop system as well.

I've also acquired a couple Magma "split-bridge" cards that may be able to be used for a PCI expansion setup but I don't have a cable to try it with. If I do find a cable, I'm going to test with some low-end stuff since it is unclear if these cards can just be hooked together or if these are just host cards. I wasn't able to find any documentation whatsoever on them. Maybe I will try e-mailing onestopsystems to see if they have info on them anymore.
User avatar
cyclone3d
l33t
 
Posts: 3038
Joined: 2015-4-08 @ 06:06
Location: Huntsville, AL USA

Re: PCie-to PCI, PCI to ISA, # of slots multipliers - bridges, risers, backplanes, research, especially for DOS, WIP.

Postby LSS10999 » 2019-5-09 @ 03:36

Doomn00b wrote:EXCELLENT find, my dude!

So, there might be a form of virtualization available for legacy devices, eh? We just have to figure out more about it, lol! Custom Bios's are a thing, so once we know more, we shalt not let that stop us.

In the meantime, I've been checking out various AMD 790FX and 790GX bords, and a few look quite promising! One actually has 7 slots, and would hence be suitable - I believe it takes at least one AM3+ cpu as well...


MSI 790FX-GD70

DFI DK 790FXB-M3H5


I'm still looking at it, as I'm not really experienced in low-level assembly programming I'm still having difficulty figuring out the correct values to use (in assembly) to communicate with the chipset registers (it might be doable with simple DOS programs as you can directly interact with the low-level hardware there)... but anyway, what you've mentioned was definitely something interesting to look into.

The Adlib/MIDI/Audio (and a lot of other legacy stuffs) SMI# trigger settings are inside Power Management registers... Not sure about the extent of their usefulness... maybe it's just to specify which legacy address can trigger SMI#.

I think the BIOS should already have SMM handlers for things like keyboard/mouse so that USB ones could behave the same as PS/2 ones in legacy environments, but as for Adlib/MIDI/Audio SMM handlers... it's still a mystery about how it should work, whether it's just to emulate the signals TSRs needed so they can work, or it needs to completely emulate a legacy sound card and send the result to a target sound card (so it can be heard).

EDIT: Was able to set up a testbed using a M5A78L/USB3 (760G/SB710, and an FX processor for confirming Vishera CPU support). By default, the PCI audio cards I tested can be detected by HWiNFO (which confirms its presence), but TSRs either couldn't detect (ES1370), or it could detect and load (AU8830), but programs still cannot detect the presence of Sound Blaster. Also, for the former (ES1370), even MPXPlay couldn't pick it up despite it's supposed to be supported natively. The onboard Intel HDA codec, however, can be picked up by MPXPlay.

Currently it's only some basic testing. It's apparent that existing PCI audio card TSRs don't work at all (not even FM Synth) with AMD 700 series (as well as future series). For further tests, I'll need to take some time figuring out more about how to access the SMBUS registers to take a peek, or find a good tool that could display chipset register settings inside a BIOS image (so one can modify it to enable the bits that we need, to test further).

EDIT 2 (CPU support related, off-topic): Regarding FX processor support, unfortunately, only the lesser 760G/770 (with SB710) chipset boards from some manufacturers have newer PCB revisions and official BIOS support for FX processors. There are no known boards with high-end 790FX/GX chipset (with SB750) that received official support for FX series (although it might be doable through possible BIOS mods I'm yet to actually find one).

EDIT 3: I was able to find the same registers in the SB600 register reference guide as well, so the possibility has existed for a while, and it was on the same address location. However, in SB800 register reference guide the particular registers were removed (Reserved).

EDIT 4: Tried a SB Live! 5.1... SBEINIT loads (taking about 4K) but programs can't see it. However, this card can be picked up natively by MPXPLAY and audio works correctly there (so native PCI access works). In such case, the SB710's onboard Intel HD should also work if accessed natively. It's now just how to peek the registers (maybe with PCISET) to see whether the necessary signals are enabled, and whether it's possible to enable legacy audio related SMI# triggers...
LSS10999
Member
 
Posts: 150
Joined: 2009-10-28 @ 14:32

Re: PCie-to PCI, PCI to ISA, # of slots multipliers - bridges, risers, backplanes, research, especially for DOS, WIP.

Postby Kamerat » 2019-5-16 @ 07:33

Have gotten myself a PCIe to PCI bridge based on the PEX 8112 chip that I'll be playing around with.
User avatar
Kamerat
Oldbie
 
Posts: 914
Joined: 2014-3-14 @ 19:09
Location: Oslo, Norway

Re: PCie-to PCI, PCI to ISA, # of slots multipliers - bridges, risers, backplanes, research, especially for DOS, WIP.

Postby ruthan » 2019-5-16 @ 12:12

Kamerat wrote:Have gotten myself a PCIe to PCI bridge based on the PEX 8112 chip that I'll be playing around with.

What are expectations, there is something better in specs than in Pericom / ASM sheets?
https://www.macrogroup.ru/sites/default ... 111207.pdf
Im old goal oriented goatman, i care about facts and freedom, not about egos+prejudices. Hoarding=sickness. If you want respect, gain it by your behavior. I hate stupid SW limits, SW=virtual world, everything should be possible if you have enough HW.
User avatar
ruthan
Oldbie
 
Posts: 1016
Joined: 2013-3-07 @ 04:01
Location: Schwarz Wald-from France to Ukraine, from Denmark to Austria. Celts+German+Slavs melting pot.

Re: PCie-to PCI, PCI to ISA, # of slots multipliers - bridges, risers, backplanes, research, especially for DOS, WIP.

Postby cyclone3d » 2019-5-16 @ 14:53

I have a development board that use a PEX chip. It has either 3 or 4 PCI slots. Has some different jumpers and LEDs on it as well.

I haven't tested anything with it yet though. I just got it because it was a good price and knew I would probably never see one again.

My experience with some other PCIe to PCI adapters has been that they work great with some cards and not at all with other cards.

Fun fact. There are a couple different PEX chips. They are what pretty much all the newer-ish PCIe cards that were back-ported to PCI use.
User avatar
cyclone3d
l33t
 
Posts: 3038
Joined: 2015-4-08 @ 06:06
Location: Huntsville, AL USA

Previous

Return to General Old Hardware

Who is online

Users browsing this forum: alvaro84, cskamacska, Deksor, Google [Bot], henryVK, Jager, knowledge [bot] and 8 guests