What about this wikipedia info:

The original write-through versions of the chip are marked with a laser-embossed “&E,” while the write-back-enabled versions are marked “&EW.”

Any Speedsys cache score differences with both jumper positions?

hard to see but i cant see &E or &EW or anything with WB lasered on the chip.
But SK096 should be wb as wiki say..

Did not bench for now i just setup jumpers for P24D with WB and look if it boots, and yes it boots fine with bios set to WB.

Then I think you're alright. (assumption, benchmark for numbers 😉 )

The P24D jumper setting takes care of the WB setting, and the internal clock multiplier of the DX4 does the rest.

If it's true SK096 then it also has 16KB of L1 - check with Speedsys or checkcpu

Ok will do the benchs later.
I also set the multi to 2 to get 66MHz. I think its nice to have the option switch between 100 and 66 setup . The idea is running on 66Mhz with attached turbo button for switching it down for speed sensetive games. With no L1/L2 i should reach a 386/33 region hopefully.
And adding a extra switch to the Multi header for switching to 100Mhz for the later stuff.
My X5 was to fast even with no L1/L2 and base clock 133 for games like Ultima7..

Jackhead wrote on 2024-08-26, 09:05:

hard to see but i cant see &E or &EW or anything with WB lasered on the chip. But SK096 should be wb as wiki say.. […]
Show full quote

hard to see but i cant see &E or &EW or anything with WB lasered on the chip.
But SK096 should be wb as wiki say..

The attachment IMG_7230.jpeg is no longer available

Did not bench for now i just setup jumpers for P24D with WB and look if it boots, and yes it boots fine with bios set to WB.

&EW is easy to see on your picture.
So it is a write back model.

Looks good:

How do i get out of cpu protection mode again? For benching the cache.

Run it under pure DOS.
No EMM386.

(Probably boot with F8)

Perhaps you try ctcm from heise Magazin too.
https://ftp.heise.de/ct/ctsi/ctcm17a.zip

Press Shift+F5 at boot to bypass config.sys and autoexec.bat.

Here the result, normal for the DX4 ?

and here with slowest hardware settings. Really slow 😀 I think that could be 386/33 range.

386DX-40 has score around 8. So this is around 386DX-20. Too slow 😉

Jackhead wrote on 2024-08-26, 16:50:

Here the result, normal for the DX4 ?

I don't have had any Intel DX4 yet.
But your speedsys cache/memory chart looks quite okay for 33 MHz FSB.
Can you test your system for stability?
(Play different games with sound more than 15 minutes, like Raptor, Duke3D, Doom - try to install Windows NT 4.o)

Ok did some testing and with the jumper settings "P24D" Whatever D or CT difference is. I know it has to do something with the codename. But i cant find information about it.
Anyway with P24D i had strange issues with crashes loading software. System boots fine even benchmark np.
So i crosstest anything again. Also switched the I/O controller, nothing.
Now i jumped to "P24CT" I found intel DX4 is codename P24C but what the is than the T again?...
It works now.

I have also a sheet where complete different jumper setup:

Oh another Intel DX4 "ODP" ?? Come on... 😀

The Asus Manual is pointing on REV 1.x i think, and the sheet is Rev 2.0 2.1.
This is really creazy stuff.

Here the Bench with P24CT jumped

You can notice that the CPU name has changed, and the L1 cache is slower. 1point more on cpu bench.
L1 is in WT mode, but jumped to WB and bios is also set to WB.
Wonder what jumper is needed to get WB. As i say JP21 is jumped 3-4.

For an Intel DX4-100 with L1 WB you need to follow the settings labeled i486DX4-&EW in the jumper sheet above. It should be the same settings as an AMD 5x86 but without setting the 2x-4x multiplier jumper JP20. Don't forget to also set the proper Hardware Trap jumpers JP5 and JP6 to 1-2.

The jumper that toggles L1 between WT and WB is JP21 (off for WB, set to 2-3 for WT).

As for the 486 CPU codenames:

• P4: 486DX
• P23: 486SX
• P24: 486DX2
• P24C: 486DX4 (same codename for both L1 WT and WB versions)
• P24D: 486DX2-WB
• P24T: Pentium Overdrive

The suffix "-S" indicates an SL-enhanced CPU (power-saving capabilities).

As for the P24CT, I think this was supposed to be a 3.3V Pentium Overdrive that was planned but never released, so it may have ended up in the manual because it was written before Intel decided not to go forward with that CPU.

ah great Info thanks!

JP5 6 are 1-2 short

here my cpu jumper:

Ends up in no boot.

Jackhead wrote on 2024-08-28, 19:15:

ah great Info thanks! […]
Show full quote

ah great Info thanks!

JP5 6 are 1-2 short

here my cpu jumper:

The attachment IMG_7243.jpeg is no longer available

Ends up in no boot.

Hmm.
You had before the Am5x86 in L1 WB mode. Is it correct?
Then just use the exact same setting for the Intel DX4-WB.
The setting in Bios for L1 WB or WT does not matter at all, you need correct jumper settings to run L1 in WB mode (for Intel or AMD 5x86 CPUs).

EDIT:
Your jumper setting is bad. Hopefully you did not fry the CPU, because you forced 5 volts by disabling voltage auto detection. Always set a jumper to JP16 1-2.

Thanks for the info. I think my CPU is still alive..
I found out that with that settings the system only boot when i disable L1 or l2 in bios.? Both enable dont work.
And yes i have jp16 now 1-2 shorted 😀

And yes i run the x5 with WB L1/2 . I also have that bios with L2 WB fix installed.

Another try, so i removed my MB and double check anything under the scope. Anything looks good.
So i installed only the VGA to test system starts. But again with L1 and L2 no boot. When i turn off L1 or L2 system boot.
Atm i think my board are maybe die..
Would be great if anyone with the board and a DX4 can tell me my jumper are correct.