https://imgur.com/a/DsSlxHE

OK, here are some photos of the motherboard.
Does anyone notice anything obviously wrong with it?
There is a bent pin on the 4th ISA slot; I have just avoided using it.

The first thing I want to do is use a modern ATX power supply (with an adapter) to rule out the PSU. So far, I've been using the AT-one that came with the case. Occasionally during boot, the HDD had trouble starting up. But the HDD is otherwise perfectly fine; tested it with Scandisk and there were no errors/bad sectors found. So I'm guessing the 5V line might also be weak, causing instability?

make super sure your ram is FPM, those glitches do sound like almost working EDO
I would also remove cache just for a test

OK, thank you.
The 16 MB RAM sticks I used to test were definitely EDO.
The 2 that are now installed (and came with the PC) are FPM (just checked; they are marked as FPM).

Is it OK if i just set the jumpers for less cache instead of removing the cache chips?
I'm a bit afraid of breaking something ...

cache disabled with jumpers should suffice for a test

https://imgur.com/BIHMVlj

Progress report:

• Simply trying with an ATX PSU didn't change anything. So, it's not the PSU

• Decreasing cache size to 2 bank 128 KB and 64 KB as well as 1 bank 128 KB via jumpers: the system still crashed. With 2 bank 128 KB In Tyrian 2000 (idle mode), Instead of lasting about 5 minutes, it took about 10 minutes to crash. With 2 bank 64 KB, there was a system restart after about a minute of launching the game

• Disabling external cache in BIOS: the system is stable.

So, I guess it's time to start removing those cache chips. Any tips for that? Should I use contact cleaner before removing chips from the sockets?
And probably use a plastic tool or something?

OK, my logic is as follows:
Since I set the jumpers to only work with one bank and I still got the same error, it makes sense that the faulty chip is in bank 0.
So, I removed all chips from bank 0 (except for the tag chip) and moved the 4 chips from bank 1 to bank 0.
Testing now.
Even if I only get 128 KB to work, I'll be happy.

Update:
Changing chips between banks didn't work.
I then bought some 15ns 256k chips and inserted them today.
Before that, I cleaned the sockets with some contact spray.
Left the timings on auto.
But still, the same error appears after running Tyrian 2000 in idle for a few minutes.

Let's say I want to test the SRAM chips one by one in order to find out if any of them are faulty. How could I do that?
Try inserting each one of them as the tag chip and see if the system still boots?

Any other ideas; maybe something else is wrong with the board and even the original SRAM chips are actually OK?

eeJay wrote on 2024-02-08, 14:42:

Update: Changing chips between banks didn't work. I then bought some 15ns 256k chips and inserted them today. Before that, I cle […]
Show full quote

Update:
Changing chips between banks didn't work.
I then bought some 15ns 256k chips and inserted them today.
Before that, I cleaned the sockets with some contact spray.
Left the timings on auto.
But still, the same error appears after running Tyrian 2000 in idle for a few minutes.

Let's say I want to test the SRAM chips one by one in order to find out if any of them are faulty. How could I do that?
Try inserting each one of them as the tag chip and see if the system still boots?

Any other ideas; maybe something else is wrong with the board and even the original SRAM chips are actually OK?

With unstable srams your computer will crash very fast, not after a few minutes.
Maybe you have corrupted data on the harddisk? Because when you wrote the data, the system was unstable?
If yes, it would be best to wipe everything, by using fdisk and format and then installing everything again.

@CoffeeOne: I already tried that; went from the original HDD to an adapter with SD card + fresh install.

Another update:
I now left the demo mode of Tyrian (instead of Tyrian 2000) running; has been stable for over an hour.
Maybe it really is just an issue of corrupted software?
Like I said: Doom, Quake, Jazz Jackrabbit, even Death Rally have given me no issues.
On the other hand, Tyrian 2000, Raptor and Terminal Velocity keep crashing.

eeJay wrote on 2024-02-08, 16:42:

@CoffeeOne: I already tried that; went from the original HDD to an adapter with SD card + fresh install. […]
Show full quote

@CoffeeOne: I already tried that; went from the original HDD to an adapter with SD card + fresh install.

Another update:
I now left the demo mode of Tyrian (instead of Tyrian 2000) running; has been stable for over an hour.
Maybe it really is just an issue of corrupted software?
Like I said: Doom, Quake, Jazz Jackrabbit, even Death Rally have given me no issues.
On the other hand, Tyrian 2000, Raptor and Terminal Velocity keep crashing.

Did you install Tyrian when you had EDO RAM inside?

Tried a different IDE cable?

Did you install Tyrian when you had EDO RAM inside?

No, I set it up when the two original FP sticks were in. I copied the game files from a USB stick (I have a couple of those ISA USB cards). Maybe that caused the corruption. Oh, and I didn't actually "install" the game, just copied the files over, ran the sound setup and then the game.

Tried a different IDE cable?

No, but definitely worth a try; thank you for the tip.

Did you checked this;

SOME CACHE MEMORY CONSIDERATIONS TOP
==================================================
Since single bank cache installations (B, D, F) prevent the cache sub-system from using interleaving access, the "Auto Configuration" in the Advanced Chipset Setup must be disabled and the "Cache Read Wait State" must also be set to "3-2-2-2" to ensure proper operation.

Source;
https://www.elhvb.com/webhq/models/486vlb3/m912v17.htm

Download the manual. In memory.htm