Thanks for the manual. I also happen to own one mb with ALD chipset and PB cache.

Will wait for your findings, but II found it actually slower than other 486 PCI motherboards I own (such as PC Chips 919 or MSI4144).

IMHO pipelined burst cache organisation provides no benefit at 486 FSB speeds (25/33/40 MHz). A well optimised 486 PCI chipset, say Sis496 or UMC can achieve 2-1-1-1, L2 reads with 15ns SRAMs already. So having a pipelined burst SRAMs with 3-1-1-1 can't change anything (and is actually slightly slower due to pipeline lead time). PBSRAM was really a solution for Pentium 60 or 66 MHz FSB speeds which would not maintain fast burst timing with async SRAMs. It is just like EDO RAMs for 486/first Pentiums

It could play a role with FSB overclocking since 10ns SRAMs are pretty rare. However, my board seems max out at 40 MHz.

Thanks ! I think some other member here figured out that the ALD marked chips are either UMC or SIS depending on which markings, sorry can not find the forum link now....

Thank you both so much for this info, I made a new page for the board on TRW.
https://theretroweb.com/motherboards/s/ald-pci5433

Is it possible to have a BIOS dump for this board (either one, both ideally)?
These ALD boards are ultra-rare, anything related to them helps a lot.

Horun wrote on 2022-12-06, 02:29:

Thanks ! I think some other member here figured out that the ALD marked chips are either UMC or SIS depending on which markings, sorry can not find the forum link now....

I've been researching ALD chipsets with a few others and came to the conclusion the infamous SARC RC2016 is possibly an ALD relabel (see DAT303), there's photo proof floating around that SARC RC2016 was used on the same PCB with an ALD 93C305.

So, hearing that ALD themselves could be a relabeler sounds interesting.

I also have a PCI5433, so this manual is useful.
Unfortunately, mine has the CPU soldered down and no cache. I was thinking to solder in the cache chips, but so far my board is not stable.
My board uses electrolytic filter caps, and BJT NPN transistor instead of a MOSFET...not exactly markings of a high quality product. I haven't had time to troubleshoot, but these are the first things on my list to check out.
I highly doubt the chipset on this board is remarked. At least I'm not aware of any other single chip 486 chipsets out there.
Even if the EDO and PLB support are less than useless on a 486, it's still pretty cool to see it work. These boards were from 1997, and I suspect the move to PLB cache was for reasons other than performance...perhaps it cost less.

Does your board come with the AMI or AWARD BIOS?

There is at least one other single chip 486 design, from ADC, with a "NEC" marked ADC004 chip (I have my doubts that's a genuine NEC part).
https://theretroweb.com/motherboards/s/adc-adgw2

Unfortunately, this one is even more rare than any of the ALD boards, I don't know of anyone else besides me that has this board, with this chipset.

I have my doubts as well about ALD being a relabeler, they even had their own web page at some point.
https://web.archive.org/web/20010627211840fw_ … t/products.html

I’ve seen those ”NEC” chipset boards. I always figured they were remarked ALDs. I don’t think I’ve ever seen one with plb cache though.

I've yet to find the ALD counterpart which this one was copied from.

The one with a known datasheet (the 93C488) doesn't electrically match mine.

Do you have a photo of your NEC board?

Anonymous Coward wrote on 2022-12-20, 13:17:

Do you have a photo of your NEC board?

Yes, here it is.

Interesting. This is quite different than the one I usually see, which had a yellow PCB and looks more like the ALD board.

I was finally able to play with this board. It has a ton of settings in BIOS some of which are uncommon and not obvious if they improve performance or not. Found it faster with L2 cache in flow-through than pipeline mode (there is a jumper next to the CPU).

With some tuning I was able almost to match performance of other modern PCI boards, except for a bit slower memory speed and write speeds.

With AMD 5x86 133 MHz, PCI Trio64V+ and EDO RAM I can get 13.9fps in Quake timedemo. Which is close to typical 14.1fps of my SiS496 boards do (MSI-4144 or Asus PVI-486SP3) and a little bit slower than ~14.4fps on UM8886 board (PC Chips M919) or ALI Finalli boards (PC CHIPS 918i or MSI4145). Not too bad. Maybe a further tuning is possible.

The board seems to come with CMA8819(A) clock generator with no datasheet and documented settings for 25/33/40 MHz only. However, the same chip is on my M919 where it is known to have more settings (and one more jumper for extra combinations). So perhaps overclocking is possible (and the pipelined burst cache could show it muscles at 50/60MHz FSB).