Are you sure the chipset demands / allows 2 x 256K TAG-RAM for 512K L2 cache size?

Answer: yes, it obviously does.

It seems like L1 cache is not being detected / used in the first place!
This would explain the poor performance.

CPU jumpering might be wrong, if you´re using a DX4.

majestyk wrote on 2022-01-22, 08:01:

CPU jumpering might be wrong, if you´re using a DX4.

Thanks for your input! I switched to a DX-33, red marked the jumpers in question, and got same result. Chipset, cache controller problem?

If the AWARD BIOS has a "Reset Configuration Data" setting under "PNP/PCI Configuration" , did you try to set it to "Enable" and "Save & Exit Setup" afterwards, after you have made hardware changes like this?

majestyk wrote on 2022-01-22, 10:46:

If the AWARD BIOS has a "Reset Configuration Data" setting under "PNP/PCI Configuration" , did you try to set it to "Enable" and "Save & Exit Setup" afterwards, after you have made hardware changes like this?

Thanks again for your help. Unfortunately this option ("PNP/PCI Configuration") doesnt exist. And there is no clear cmos jumper (but I removed battery to clear cmos data). and I always loaded setup/bios defaults. I double checked all jumpers and theyre right (WT L1-Cache CPU is jumpered, WB cache inside BIOS for L2-Cache).
Seems we are stuck with that one?

How did you manage to remove the battery of a Benchmark BQ3287MT? It doesn´t even have a RAM clear input (like the BQ3287A).

majestyk wrote on 2022-01-22, 14:24:

How did you manage to remove the battery of a Benchmark BQ3287MT? It doesn´t even have a RAM clear input (like the BQ3287A).

No, just took it off and reseated it. sorry for confusion.

O.k. - but that´s not resetting the stored data.
Do you have a RTC-clock from a different mainboard that you can insert temporarily?
It also wouldn´t hurt to reflash the BIOS - just to rule everything out except hardware...

majestyk wrote on 2022-01-22, 16:08:

It also wouldn´t hurt to reflash the BIOS - just to rule everything out except hardware...

Thanks once again for your support. I was able to change the RTC(Screenshot "defaults loaded") and reflash the BIOS (even different versions 0204 and 0205). Same outcome.

One more observation: On all pictures of revision 1.6 there´s a soldered jumper "1-2" in "JP27.

On your board this jumper is missing, there´s no solder in the via just as if someone had removed the jumper.

There is a soldered metal bridge between JP27 1+2 on the opposite side of the board. There is continuity between 1-2. But good observation. Unfortunately it seems not related.
What could be the defective part in such a case? Chipset? Is there something like a "common" cache controller chip? I mean L1+L2 caches are not working.

Bridging on the backside is quite unusual for ASUS boards. But we don´t know all about the past life of this board 😉

After you reflashed the BIOS and even replaced the RTC, I would rule out any software related cause at the moment. But enabling L1 at startup with the related register might be prevented by other circumstances.
The "EX" northbridge has it´s own CPU-interface. There could be some loose pin at the northbridge or in the CPU socket, a defective buffer or even a problem inside the chip.

Have you made sure that the turbo button is set correctly?

aitotat wrote on 2022-01-23, 10:39:

Have you made sure that the turbo button is set correctly?

Well... this kills me since:
- I always thought that Turbo SW is enabled by default and putting a jumper on it disables turbo mode
- Turbo mode is unrelated to cache and its just about CPU frequency.

Long story short: After putting a jumper on turbo sw, the cache showed up and the machine turned into a monster (~38 in speedsys).
I am shocked.

Thank you guys for getting this situation fixed. Beers are on me if you told me your address. 😀