Loosen your memory timings... Set both memory and cache timings to the lowest setting possible. If the freezing goes away, then the issue is that your timings are too tight.

mockingbird wrote on 2024-01-05, 03:13:

Loosen your memory timings... Set both memory and cache timings to the lowest setting possible. If the freezing goes away, then the issue is that your timings are too tight.

Thank you for your answer, I have a question,
The 486DX2-66 and X5-133 use the same 33MHz FSB, do they need to have different memory and cache settings?

The problem most likely has something to do with the VRM becuase the DX2-66 runs straight off the 5V rail provided by the PSU, and that's why it works.
Normally, I would advise you to change the caps in this situation, but they seem to be good brands (Rubycon and Elna judging by the vents).

I am certainly not a fan of recapping without proper diagnostics, but it might be worth a shot. Otherwise, you would need an oscilloscope to measure the 3.3 V output under load.

zwrr wrote on 2024-01-05, 03:45:

Thank you for your answer, I have a question,
The 486DX2-66 and X5-133 use the same 33MHz FSB, do they need to have different memory and cache settings?

I don't fully understand the mechanism of why it would matter, but try it anyways and see if it solves the issue. I had a similar issue and fixed it this way. I used Doom as the ultimate stability tester... I could play for a while and then it would lock up. Loosening my memory timings fixed it.

mockingbird wrote on 2024-01-05, 13:15:

zwrr wrote on 2024-01-05, 03:45:

Thank you for your answer, I have a question,
The 486DX2-66 and X5-133 use the same 33MHz FSB, do they need to have different memory and cache settings?

I don't fully understand the mechanism of why it would matter, but try it anyways and see if it solves the issue. I had a similar issue and fixed it this way. I used Doom as the ultimate stability tester... I could play for a while and then it would lock up. Loosening my memory timings fixed it.

The higher internal speed (133 vs 66) and the doubled L1 cache will create more accesses to the L2 cache, so it is possible.
But I agree it is the same 33MHz bus ..... So maybe there are also instabilites with 66, just the likelihood is smaller.

I fixed it by replacing the CPU socket and now everything is fine.
Need help replacing the pins of the Socket3 socket

I get the impression that the CPU socket contacts are not in good contact due to rust, which makes it look normal under DX2-66 because of the 5V voltage and lower data transfer speed, but when changed to X5-133, the 3.45V voltage and double the data transmission make it unstable.