the change in multiplier is made via jumpers on the motherboard.

The dx4 is 3x, and most certanly will work in 2x..

Check your motherboard manual, or markings in the board itself to figure it out where the multiplier jumpers are and how to set them.

do you know what model of motherboard you have ?

it is an IBM PC330 DX4

I do not see the indication of the group of jumpers used to change the multiplier printed (like all ibm) in the sheet inside the case. Could it be an undocumented function of some jumpers ??

According to http://ps-2.kev009.com/pcpartnerinfo/ctstips/a5a2.htm, the multiplier is controlled by jumper JP10. Short pins 3-4 for x2, leave it completely open for x3.

The multiplier can be changed as long as it's not a DX4 Overdrive.

ok, something happens ...

closing the JP10 CHKCPU sees the processor as 22Mhz x3 = 66Mhz and speedsys detects a CPU value of 28 and instead with the DX4 detects 39 so it is less fast. But in any case the values are strange ... will there be some other jumper to modify or it works at 33x2 and the software to detect it is wrong maybe because of the bios?

Second question... I found some cache chips and installed them like this ...

https://drive.google.com/file/d/1UGmMM3117ykX … iew?usp=sharing

is that correct? The chip of TAG is identical to the others and those nice group of 256 have a longer socket ...

I have asked a similar question some time ago.
There were a number of particularly helpful responses, explaining the different multiplier behavior on different CPUs.

between 66mhz and 100mhz (well 99mhz) the bus is the same.. 33mhz..

So 33x2=66hz, 33x3=99mhz.. thats that simple..

The processor dont scale exactly the same, and a dx4 is not certanly 1/3 faster than a dx2... or 3 times faster dan a dx-33 in a linear way... it all dependes on software and other factors like cache.

retardware wrote on 2021-11-02, 16:53:

I have asked a similar question some time ago.
There were a number of particularly helpful responses, explaining the different multiplier behavior on different CPUs.

Very interesting!!!! tnks!

nuno14272 wrote on 2021-11-02, 17:26:

between 66mhz and 100mhz (well 99mhz) the bus is the same.. 33mhz..

So 33x2=66hz, 33x3=99mhz.. thats that simple..

The processor dont scale exactly the same, and a dx4 is not certanly 1/3 faster than a dx2... or 3 times faster dan a dx-33 in a linear way... it all dependes on software and other factors like cache.

I know, but the computer tell me 22x3 instead of 33x2 seems to know the target (66mhz), the multiplier (3x instead of 2x thinking dx4 cpu) and it adapt the bus "ideally" to obtain this equation.

AlessandroB wrote on 2021-11-02, 20:32:

I know, but the computer tell me 22x3 instead of 33x2 seems to know the target (66mhz), the multiplier (3x instead of 2x thinking dx4 cpu) and it adapt the bus "ideally" to obtain this equation.

Is it possible to rule out that the strange "22 MHz" is maybe a display bug of CHKCPU, thinking it is a x3 processor, while actually it has a 33MHz bus and the multiplier input pin configures it to work at x2?

If your CPU is Cyrix/TI/IBM/ST, then it's possible that the speed is being misreported if the program in question isn't designed to properly identify it. Speedsys for example detects my 40MHz 486SXL as 13MHz despite the fact it runs at 40.

retardware wrote on 2021-11-02, 21:05:

AlessandroB wrote on 2021-11-02, 20:32:

I know, but the computer tell me 22x3 instead of 33x2 seems to know the target (66mhz), the multiplier (3x instead of 2x thinking dx4 cpu) and it adapt the bus "ideally" to obtain this equation.

Is it possible to rule out that the strange "22 MHz" is maybe a display bug of CHKCPU, thinking it is a x3 processor, while actually it has a 33MHz bus and the multiplier input pin configures it to work at x2?

yes actually is set to 33x2 and the speedsys report a corre t speed for a 486dx2

Anonymous Coward wrote on 2021-11-03, 05:07:

If your CPU is Cyrix/TI/IBM/ST, then it's possible that the speed is being misreported if the program in question isn't designed to properly identify it. Speedsys for example detects my 40MHz 486SXL as 13MHz despite the fact it runs at 40.

no it is the genuine dx4 intel installed by imb in the 1995 in this pc

Alternatively you can just set the FSB to 25MHz, a DX4-75 is roughly equivalent in performance to a DX2-66 - one or the other comes up on top with marginal difference depending on cache dependence in the benchmark.

i want to have genuine system if possible

AlessandroB wrote on 2021-11-03, 11:44:

i want to have genuine system if possible

Running a DX4-100 as a DX4-75 is as genuine as running it as a DX2-66, no idea what you mean here. Unless you have a performance target why are you doing this?

Because the 486DX266 was my first PC (after Amiga500) and I would like the feelings that from this computer to be the same as mine gave me in 1993. Just seeing the 486DX2 66 script at boot is an emotion . But only for this specific cpu, then the others if needed for speed sensitive software I can get them with an SS7 that I quickly assembled in recent days, when I realized that this DX2 is a bit too slow to be used well beyond of the memory effect

I see. Then you may as well just buy a DX2-66 CPU, they are fairly abundant still and you may regret not getting one down the line.

I have at least 7/8 cpu dx2 66 from different brands, from intel to texas instruments, from amd to ibm. I would like to have a dx4 that with the simple change of a jumper transforms it into my beloved dx2, so into a single machine without having to open it but bringing only a switch out of the case. Without forgetting that a 3v dx4 operated at 66mhz will expire even less. If I really don't succeed I will replace the dx4 with a dx2. A little above I posted the photo of the cache that I installed. Can you tell me if it is installed correctly? I all have identical chips, even the TAG one.