BIOS chip marked:
"386DX C1216
(C) 1989 AMI All Rights Reserved
386DX BIOS (Ver. 2.11)"

Chipset has the PC Chips
"CHIP 5": 4L04F1282, Japan 9306EAI
"CHIP 6": 4L04F1666, Japan 9304EAI

I'll update with the BIOS ID string just in case the markings on the BIOS chip don't reflect the actual BIOS.

I rather love the board for having a socketed tag RAM chip for the cache so upgrading to 256KB of cache was a breeze.

Do you have an eeprom burner? You may try Mr BIOS

AlaricD wrote on 2023-01-24, 18:14:

or if Address 11 or Data 7 are not connected somewhere

That's actually quite likely. Support for A11 on mainstream consumer chipset typically appeared on second-generation 486 chipsets (or 386/486 comibination chipsets). Older ones do not support 16MB SIMMs. You can easily test that: If A11 is driven by the chipset, you will encounter some parasitic diodes or ESD protection diodes. Do a diode check with a multimeter (the cheapest 5$ Aliexpress / Poundland / whatever model will do), putting red at A11 and black at +5V, and a second check putting black at A11 and red at GND. If neither of those measurements shows some conductivity, A11 is unconnected. The test is performed with "reverse polarity" intentionally, to forward-bias ESD protection/clamping diodes.

Am386DX-40 wrote on 2023-01-24, 21:50:

Do you have an eeprom burner? You may try Mr BIOS

That's another idea. I've also found some additional M321 BIOS .BIN files. I *might* be able to scrounge up an EEPROM burner-- I have a friend with one but he hasn't seen it in ages so it's probably buried under anything/everything right now.

I wonder, though, if there's a simple electrical test for continuity to ensure the address lines are all connected somewhere. Sadly, I don't have a working oscilloscope, but maybe a logic probe (again, at my friend's house) could be useful for that.

Another thing I want to fix is the flush input for the Cx486DLC-40 or TI486SXL-40 (the PGA132 version) since that may be the fastest and most-reliable cache invalidation method.

mkarcher wrote on 2023-01-24, 22:18:

AlaricD wrote on 2023-01-24, 18:14:

or if Address 11 or Data 7 are not connected somewhere

That's actually quite likely...
Do a diode check with a multimeter (the cheapest 5$ Aliexpress / Poundland / whatever model will do), putting red at A11 and black at +5V, and a second check putting black at A11 and red at GND. If neither of those measurements shows some conductivity, A11 is unconnected. The test is performed with "reverse polarity" intentionally, to forward-bias ESD protection/clamping diodes.

NICE! I'll have to try that. I probably should try to fix my multimeter-- the mode select button has quit working which might be the conductive graphite pad being worn, but maybe a new multimeter would just be the better route. Good tip on the polarity.