Adding new info to my old post as I finally got around to installing and testing this card.

The PCI-AS2940UW is a MediaBus 2.0 card which replaced the similar 1.2 version named PCI-AS7870. It also changed the SCSI chip from 7870 to 7880P. From pictures I've found there are later revisions of the AS2940UW with no OPL3 and the CT2505 Vibra chipset (instead of CT2504 as installed on my card). At some point the card was renamed again as the PCI-AS300, but with the same p/n as the AS2940UW. According to a source in Google Groups this was the longest selling MediaBus card offered by Asus.

I've installed it into my P65UP5 rev 1.41 board and it's not been a pleasant experience. The SCSI part works without issue, but with a Matrox Millennium II installed the computer hard freezes once the sound is enabled. DOS works fine, crashes only happen in Windows 95. So far I've tried the following:

- Get latest drivers for Matrox and Soundblaster
- Manually assign IRQ/DMA for ISA part and PCI slots in BIOS.
- Change IRQ/DMA for both Matrox, SCSI and Soundblaster (no conflicts).
- Reinstall Windows95

I plan to try and disable busmastering on the Matrox (with performance hit), play around with BIOS PCI settings and update the BIOS on the Matrox card. Currently the Matrox is installed in Slot1 and the MediaBus card in Slot 5. Slot1 initializes first and only slot 4 and 5 shares IRQ. I'll also try to swap position to slot2.

If everything fails I'll get a new SCSI card (AH2940U2W) and use a separate AWE64.

Basically it's as simple as a PCI SCSI card and an ISA (PnP) sound card on the same PCB. What you're hitting here is good old ISA PnP hell, which you almost certainly would have had with a similar Vibra on ISA card too.

dionb wrote on 2020-08-31, 09:30:

Basically it's as simple as a PCI SCSI card and an ISA (PnP) sound card on the same PCB. What you're hitting here is good old ISA PnP hell, which you almost certainly would have had with a similar Vibra on ISA card too.

I had the AWE64 in the machine before using the Vibra16 part of the PCI-AS2940UW, the sound worked with no freezes then. The Vibra16 is a semi pnp card and might play a part, but it could also be due to a quirk in how the MediaBus is implemented on the card/board.

I've done more researching into the matter and I've finally gotten to the bottom of it. The issue is not with the MediaBus, but the combination Intel 440FX, Matrox Millennium and the Soundblaster 16.

Before I found the solution I tried the following:

- Updating PCI IDE Bus mastering chipset drivers to latest version
- Update Matrox Millennium video BIOS from v1.0 to v1.5
- Add [mga.drv]PCIChipset=1 to system.ini file
- Disable busmastering on the Matrox Millennium
- Update to the latest Win95 driver for the Adaptec A7880 chipset.

None worked. I then went into the BIOS and just started disabling chipset PCI write options. When I got to "USCW write posting" i finally struck gold. This needs to be disabled.

I then looked into the matter more and found out I haven't been the only one with this issue. A thread from 96/97 at Google Groups had the following info:

USWC Write Post During I/O Bridge Access Enable (UWPIO). 1=Enable. < allows posting of CPU-to-PCI cycles destined for a USWC (Un […]
Show full quote

USWC Write Post During I/O Bridge Access Enable (UWPIO). 1=Enable.
<
allows posting of CPU-to-PCI cycles destined for a USWC (Uncacheable,
speculative, write combining; essentially PCI adapter memory for which
writes to can be combined/cached, like an SVGA adapter) region, even during
a passive release cycle.

It says nothing specific about what passive release is, other than:

To comply with PCI Specification, revision 2.1 latency requirements, the
PMC supports passive release. The PMC disables CPU-to-PCI posting during
the passive release, except for transactions to the USWC region (resumably
only if the above bit is turned on).

...

I saw another post which seemed to imply this conflict is only caused on
machines with a Matrox Millenium and a Sound Blaster, and that the actual hang
occurs when the Sound Blaster is initialized / used. This would be consistent
withit hanging on my system as Win95 starts (initializing the SB), and on your
system when running Quake in DOS (using the SB).

I finally found a technical summary document (http://www.ocean.aust.com/tech.html) that had some information about what passive […]
Show full quote

I finally found a technical summary document
(http://www.ocean.aust.com/tech.html) that had some information about what
passive release is. Supposedly "This mechanism enables more efficient ISA Bus
Master cycles to PCI by absorbing ISA overhead before requesting PCI. On
multiple data cycles, the PCI bus is free for other agents' use in between
each ISA bus data phase." It goes on to say "As an example, audio cards place
less load on the system, enabling more CPU and PCI concurrency."

I guess this explains why the SB is the triggering agent for the failure,
since it would be the only thing (in my system, it is the only ISA card) which
could cause a passive release cycle. It also explains why FASTVID (or
the BIOS setting to enable write combining) reveals the problem, since
without it the LFB or VGA buffer are not treated as USWC regions, but are
instead just plain uncacheable memory. I imagine it means the only
performance aspect affected by this setting is writing to the SVGA card
concurrent with using the SB on a system with FASTVID write combining options
enabled. This would probably be a factor for most games (how much is
anybody's guess; only someone w/o this bug could tell us), but explains why
the setting does not seem to affect VSPEED, and why the setting seems to do
nothing for a system without FASTVID write combining turned on.

Of course, why it fails is anybody's guess. Probably some bug in the Matrox
card handling writes during a passive release cycle, or perhaps the
440FX chipset itself.

What is strange is that I didnt have any issues with the AWE64, meaning it could not be affected/triggering this bug. I also don't know how much performance will be hit here since FASTVID can still circumvent this according to the thread. I'll post some benchmarks later. Please also note that "Video Cache Mode" can still be set to USWC in BIOS without issues.

TLDR: Bug in either 440FX chipset or Matrox Millennium when used in combination with SB16 on "passive release" with USWC write posting BIOS option.

Hello,
I had gotten one of those Asus PCI-AS300 is a lot box not to long ago. I have Asus P55T2P4, I know it has a different chipset, 430HX.
I was also going to use a Matrox Millennium. Once i get the system together, I'll update if i run into a similar issues or not.

Asmodeusbell wrote on 2021-03-03, 23:29:

Hello,
I had gotten one of those Asus PCI-AS300 is a lot box not to long ago. I have Asus P55T2P4, I know it has a different chipset, 430HX.
I was also going to use a Matrox Millennium. Once i get the system together, I'll update if i run into a similar issues or not.

Please let me know how it goes. Afaik AS300 is an even later revision, main differences is that the OPL3 chip is replaced with CQM.

Hi
I have Asus P55T2P4 and have bought Asus PCI-AS300 recently. Sound card works like standard Sound Blaster but I can't make SCSI work. When computer starts (post screen) it shows only pci sound card on list and then it doesn't show SCSI BIOS, hdd led doesn't light up and SCSI HDD doesn't show up in DOS. Any advise?

Maybe the Adaptec BIOS is bad ? Because the SCSI runs only the PCI part and the Vibra 16C runs only on the Media Bus 2.0 you can try the card in a different PCI slot and it (SCSI) should show up.
If you have a TL866 you should try to read the BIOS chip to see if it is good, pictures of AS-300 show it as a removable ROM chip.

Horun wrote on 2021-08-21, 17:02:

Maybe the Adaptec BIOS is bad ? Because the SCSI runs only the PCI part and the Vibra 16C runs only on the Media Bus 2.0 you can try the card in a different PCI slot and it (SCSI) should show up.
If you have a TL866 you should try to read the BIOS chip to see if it is good, pictures of AS-300 show it as a removable ROM chip.

Thanks for reply
I have tried different PCI slot and it didn't show up.
I have read SCSI BIOS without problems and programmed it with a file I found on Internet. No change.
This card is new, never used, so could be defective from factory 🙁

Do you have another motherboard you can test the card in? As Horun mentioned, only the Vibra16 part required the MediaBus.

I have tried it on other motherboard. It doesn't show up during boot..

May not help but suggest you clean the PCI slot connector portion on your card gently with pink eraser and then paper towel with IPA and try again.
There might be residule manufacture oils or laquers on those tiny fingers 😀

Cleaning PCI connector didn't help.
I have exchanged few messages with the seller. He said that this is first time someone has problem with this card and he has sold a lot of them.
Last message from him is:

This card only works in MediaBus slot so don't use in PCI. The issue is with your motheboard. Please try in another Asus motherboard with a media bus slot.

I think that I might have another board with MediaBus..

Man that's weird, I have/had several T2P4s and I seem to have blanked the existence of the media bus on them from my mind... maybe I ignored it because I thought it was useless.

I have just checked card on another Asus MediaBus motherboard (TX97) and it doesn't work..

Would it be possible to obtain BIOS dump from the SCSI part? I bought this card locally but some genius has pulled the BIOS chip on it and of course no one knows where it ended up (most likely in scrappers)..

I have for Asus PCI-AS300, one is from my card and another was found on internet

Ok, thanks! I will try it even though I have PCI-AS2940UW.

Also, what BIOS chip is there? I would have to find a suitable one regarding leg count, voltage, size etc.