kotel wrote on 2025-11-09, 16:57:

Hi […]
Show full quote

Hi

Recently I got yet another AMD socket A cpu (Athlon XP 2200+, orange plastic PCB) from scrap. It had a few bent pins which I've since straightend and soldered one since it broke off. The die had a few little chips in the corners, but no large chunks missing. I then inserted it into a KT400 based ASRock K7VT4A and powered on.

Sadly, all I got on the POST card were dashes (no POST code execution) and IRDY with FRAME LEDs being lit.
Just for a sanity check I've swapped in the previous known good 750mHz Spitfire Duron and everything worked fine.

I then checked if the die was getting warm. The only part that did get hot was the bottom one. Top one was cold.
Another forum poster told me to cut out the jumpers for the cache since it might just work. Sadly, that didn't change anything.

This wouldn't be so weird if not for the fact that this is the 3rd plastic AMD CPU which did this for me. The two other were XP 1700+ (green) and 2200+ (orange) with pretty much no chips to the die tested on the same mainboard (and a known good Asus A7V8X-VM). The 1700+ burnt itself when I tried testing it (noticable pop sound and burnt plastic smell from the CPU). There also is a 800mHz Spitfire Duron too, although this one makes the VRM squeal loudly (with IRDY and FRAME LEDs lit).

Also tried them in an KT333 based MSI KT3 Ultra 2 but with same results.

I do account that there might be micro cracks in the bent pins on the chiped 2200+ and even in the die, but on the 1700+ and 2200+ which were in pretty much perfect condition? Don't think so.
I have never seen such a thing, especially the 1700+ burning up.

Anybody else having such issues with socket A CPUs?

Several years ago I tried various AMD CPUs on a PCChips SiS chipset (equal to ECS), however several were not working, most were Athlon XP, maybe they need a very large and efficient heatsink, if you used a heatsink for S.370.
I also remember an Athlon 800 that heated up a lot, the same CPU in another PCChips motherboard, instead it works well 😌, it's nice and cool.
Maybe it could be something in the VRM, which creates problems and heats up the CPUs faster, or the high-frequency ones (?).

Back when these CPUs were current, chips on the corners, even if very small could kill them.

I had quite a few that I tested a few years ago and barely any of them worked. I think it was a out 10 I tested and only 1 or 2 were functional.

Other than verifying that you are using other known good components, there isn't really anything you can do.

A chipped corner wouldn't normally kill the cpu, it would be cracks inside that reached the bottom. I have seen some badly chipped cpus, but probably because low force was used when chipping they still worked fine.

So I did a quick test with IPA on the XP 2200+ die.
The core arena outlined in red evaporates IPA the fastest (see attachment). I do not know if that's the main CPU core or the cache, so I'd like to ask somebody to identify which part it actually is.

Both VRMs of the mainboards are fully working (boards after recapping).
I've tried a technique using a thin coat of IPA, letting the die soak it up and then pressing on the die. No cracks appeared when doing this (or maybe I just did this wrong on my XP 2200+).