All I can find as nearest looking card is PROMISE DC420 - https://theretroweb.com/expansioncards/s/promise-dc420
BIOS - https://www.vogonsdrivers.com/getfile.php?fileid=2155

Thank you very much. I had also seen those links and tried a couple of things, but without success.

I'd need to be able to extract the BIOS and compare it with the ones provided for the DC420, maybe the current BIOS is corrupted.

And I don't know if adding a BIOS chip would do anything. This card is so mysterious, even the OAsys chip is strange.

OAsys gives me NEC CMOS-8 gate array vibes, meaning its custom whatever they needed inside

It looks very similar to the guy in this thread.
Help to identify this VLB io hdd cache controller

Can OP compare?

Paldiplex wrote on 2026-09-29, 11:38:

Thank you very much. I had also seen those links and tried a couple of things, but without success.

I'd need to be able to extract the BIOS and compare it with the ones provided for the DC420, maybe the current BIOS is corrupted.

And I don't know if adding a BIOS chip would do anything. This card is so mysterious, even the OAsys chip is strange.

There is USUAL a 2 ROM chips on such cards - one is computer BIOS extension and second is internal ROM for onboard 286 CPU
One of the chips is missing on your card if you don't see any BIOS extensions output on your screen I think its computer BIOS extensions
Maybe im wrong

On your photo ROM address jumper set to NO-ROM position just try it to other positions like C8 (C800h) and JP3 to 2-3 (HDD enable) and JP5 to 2-3 (RAM 16 - I think its for 16bit onboard RAM access 2 SIMMs mode)

wierd_w wrote on 2026-09-29, 12:18:

It looks very similar to the guy in this thread.
Help to identify this VLB io hdd cache controller

Can OP compare?

Yes, it's exactly the same, and even though I looked for it, I hadn't seen it. I feel less alone. Thank you.

And as for the other participants, I'll take a look at your suggestions and get back to you. Thanks to everyone, that's already a lot. 😀

Nothing in The RetroWeb or Total Hardware for that card. I'm curious.

With a 1993 BIOS date, it's unlikely to support drives over 512MB. Maybe you can enter them into the config screen, but they probably won't stick.

Any chance you can dump the ROM with a chip reader like a TL866Plus ? There's probably a copyright notice or something in there.

I noticed that the PCB part number matches the part number on that OASys chip.

If there is room after the main pc bios on the chip, it might be able to work with xtide configured for promise vlb, if it can fit on the chip.

From xtide build instructions page:

MODULE_ADVANCED_ATA ¶

Adds native support for VLB (and hopefully in the future also PCI) IDE controllers. There is currently support for QDI Vision QD6500 and QD6580 VLB IDE controllers. There is also experimental support for Promise PDC 20230-C and 20630 VLB IDE controllers but this needs more testing.

I'd see if the XUB Disk booter method can successfully work with this card, with that mode on.

Thanks again to everyone for the information; I find it very interesting and intriguing.

I've run a few tests, particularly on transfer rates. The cache managed by the processor seems to be working, because with Checkit I get 3,900 KB/s in its default configuration and 1,900 KB/s when I remove the memory modules.

The EIDE2300 DOS driver seems to initialize in PIO 7 mode, but the associated Windows 3.11 driver doesn’t work (32-bit disk access isn’t functional).

I’m going to try to get an EPROM programmer to extract the current BIOS and also look into XTIDE.

I’m also surprised that I can’t access the card’s settings using a key like F2, as on some Promise cards. There’s some research to be done 😀

One thing is certain: LBA mode doesn’t work, even though it does work with a VLB VLMIO V1.6 card. Indeed, 1993 is a bit early for LBA

There is a way to start xtide from a floppy.

XUBDisk - floppy disk XTIDE Universal BIOS booter

I would test that with the image configured for advanced ata.

Compare disk io speeds with that vs stock bios.

If it works, 'maybe xtide can live in the pcbios socket'

Paldiplex wrote on Today, 07:03:

Thanks again to everyone for the information; I find it very interesting and intriguing. […]
Show full quote

Thanks again to everyone for the information; I find it very interesting and intriguing.

I've run a few tests, particularly on transfer rates. The cache managed by the processor seems to be working, because with Checkit I get 3,900 KB/s in its default configuration and 1,900 KB/s when I remove the memory modules.

The EIDE2300 DOS driver seems to initialize in PIO 7 mode, but the associated Windows 3.11 driver doesn’t work (32-bit disk access isn’t functional).

I’m going to try to get an EPROM programmer to extract the current BIOS and also look into XTIDE.

I’m also surprised that I can’t access the card’s settings using a key like F2, as on some Promise cards. There’s some research to be done 😀

One thing is certain: LBA mode doesn’t work, even though it does work with a VLB VLMIO V1.6 card. Indeed, 1993 is a bit early for LBA

PIO only went up to PIO4 in practice , and with few exceptions, PIO5 never got father than the spec sheet. Was PIO7 a typo?

With a 1993 bios, LBA and ECHS are not likely to work. You are likely limitted to 64 sectors, 16 heads, and 1024 cylinders.

The rom chip that you have likely provides the firmware for the 286 chip on your board.

A bios rom would likely go in the empty socket and be the bios extension for the computer. It would provide the “f2 for setup” stuff and an int 13h handler that does faster transfer modes & 32 bit transfers from the cache to the PC. Ideally you should be seeing transfers over 10MB/s when working out of the cache.

You could try a bios chip from a DC4030VL card . Worst thing that happened is the computer doesn’t boot and you have to pull the chip.

What seems really weird to me, is that they use a 16bit CPU and 16bit memory on a 32Bit VLB... makes me wonder if windows 32bit drivers will ever work.

BitWrangler wrote on Today, 13:15:

What seems really weird to me, is that they use a 16bit CPU and 16bit memory on a 32Bit VLB... makes me wonder if windows 32bit drivers will ever work.

Let's review.

The 16 bit 286 talks to the ATA device and copies sectors into the cache via the OAsys chip. The ATA PIO protocols are 16 bit so this isn't limitting.

Next the PDC20230C chip pulls data from the cache through the OAsys chip into a small on chip buffer. I suspect that the OAsys chip presents itself like an ATA device with high throughput and potentially high latency when the sector is not in the cache. On the promise boards, you can tweak the size of the cache lines to optimize for throughput or latency. No visibility into that here though. Anyway ...

The CPU is able to pull from the PDC20230C buffer 32 bits at a time if 32bit transfers are enabled by a driver or BIOS extension, or if you have a later 486 BIOS that supports 32bit transfers natively.

Edit -- There's a performance gain here by reducing the number of transfers over VLB and the number of writes to main memory, but it's an incremental gain, not a huge gain, expect like a 10-15% improvement in most cases. It really starts to matter if the "writes to main memory" become a real source of throughput resistance for the VLB IDE caching controllers. I've seen some of the slower VLB chipsets with memory write speeds under 20MB/s look I/O bound because the driver can't write the incoming data to main memory fast enough.

Edit #2 - Most of the performance benefit reported by "32bit drivers" in the early windows world was because the CPU didn't have to drop back to real mode to do the old DOS int13h I/O. In those cases, just staying in "32 bit protected mode" while doing I/O had a noticeable improvement in OS feel and responsiveness, regardless if the driver was communicating to storage at 32bits or 16 bits.

Makes you wonder why nobody (or did they?) make an external box pluggable between ordinary IDE drive and controller 😀
Not that it would make much sense, imo all those caching controllers are inferior to having more ram in the system and SmartDrive loaded.

rasz_pl wrote on Today, 14:00:

Makes you wonder why nobody (or did they?) make an external box pluggable between ordinary IDE drive and controller 😀
Not that it would make much sense, imo all those caching controllers are inferior to having more ram in the system and SmartDrive loaded.

I test some ISA/VLB/PCI cache controllers and they are effective with good old mechanical HDDs from late 80 up to early 90 (something up to 800MB-1GB max)

Sometimes many modern users use them with fast HDD from 1995-97+ and more modern and/or CF and SSD. In such case they are do not accelerate anything

So if users rig is good old original with pair of Conner/Seagate HDDs up to 540MB its nice to add cache-controller

rasz_pl wrote on Today, 14:00:

Makes you wonder why nobody (or did they?) make an external box pluggable between ordinary IDE drive and controller 😀
Not that it would make much sense, imo all those caching controllers are inferior to having more ram in the system and SmartDrive loaded.

Pretty much yeah. When smartdrive 4.0 came out with DOS 6.0 in 1993, computers frequently had more than 1MB ram and caching controllers faced a strong headwind against a much more efficient solution. At the same time, hard drive vendors were building ever larger caches into the drives themselves, which had access to buffer things that were just not available to the host side, like an instant replay of that block that just went under the drive head but wasn't requested until after it passed, that kind of stuff. The niche for the desktop caching controller was pretty much over as fast as it came, just like VLB itself. Although in 1993, it was still super cool to imagine with friends that if you had more money, you could buy a 286 CPU to handle the I/O in your 486-66 VLB monster. Practical? No. Aspirational? Most certainly.

Caching did stick around on servers as RAID controllers for another 10-15 years before "the SAN" rendered them obsolete. Curiously, the SAN is sort of resembles the external pluggable box between commodity drives and the computer that you were day dreaming about ( with added features for self defrags, hot swappable tiered storage, volume switching between servers, MPIO load balancing, integrated snapshots and backups, QOS controls, etc, etc ).