85C should not kill a K8 CPU if the voltage wasnt way too high.
But the heat could have made the VRMs to seriously overheat and that could lead to clusterfucks.
I still think that there is a good chance the CPU is OK but thats bad news...

We do have to consider that this gear is 10 years old so anything can go poff for no good reason.

85C is 20C over spec BUT probably still not enough to kill it. That is, if it really shut down at 85C

I ran a Prescott at 130C for a day straight without realizing it this week. Still works! But it does smell rather funny. . .remember to always make sure your cpu cooler is attached properly and that the thermal warnings in the bios are enabled!

ODwilly wrote:

I ran a Prescott at 130C for a day straight without realizing it this week.

Sure? A Prescott throttles extremly at 130C, you should mention it by a very slow system (and about 130C a Prescott stops a system completly).

But Intel != AMD, Intel has a very good overheat protection. You can start a P4 systen without a cooler. The system starts and very quick it stops because of overheating, but it will not damage anything! 😎 (tested it two times! and tom's hardware also tried it with P3 and P4 but nothing was damaged)

AMD doesn't have a good overheat protection. I heard from S939-systems that died because of a damaged cooler.

My bios by default disables the intel automatic shutdown temp, which is a very bad design flaw. Heck there is an option to disable throttling as well. Yes, it was very slow. That is why I noticed the temps finally, even 190 windows updates should not take 24 hours to install

Have you checked the PSU?

check your cooler, even on the stock cooler at stock clocks an a64 running prime95 for 3 days wont hit more than about 58-61c

I suspect the motherboard. If an Athlon64 was roasting like that, the VRM FETs probably were too. Perhaps much hotter considering the AV8 doesn't even have heatsinks on them.

All that cpu intel x amd overheating made me remember from that very old tom's hardware video, ah that music 😁
https://www.youtube.com/watch?v=06MYYB9bl70

Thanks for the replies. It sounds like the CPU is probably not what's wrong, and I've screwed the motherboard instead.

It was running hot because it had old crud on the heatsink and CPU that I didn't take the time to clean properly, and I did a sloppy job applying the new paste, which also happened to be the worst imitation of the stuff I have. It was a temporary quick setup to see if everything worked. Once I saw it was working, I should have taken everything back apart and cleaned it up properly, but I didn't.

I saw it getting into the 60s while running an OS. It had previously run memtest for about 1h30mins without issue, but I don't know how hot it got doing that. Then I plugged in some different RAM and got the bright idea to leave memtest overnight. Dumb.
I'm only assuming the 85C shutoff actually worked - maybe it got hotter than that. All I know is the next day it was shut down, for one reason or another.

Oddly the board was still POSTing at that point, but after sitting a few more hours it wouldn't POST anymore. I got frustrated and had to break down the board and move it, but I'll look into it again at some point.

I'm pretty sure the PSU is still good, I checked the main PSU voltages while it was trying to POST and they were normal. The only ones I didn't check were the standby rails.
I saw about 0.25v on the Vcore output, so I think the VRM is toast. That was a snap diagnosis though - I didn't take the time to be completely sure I was looking at Vcore.
The most convenient explanation for sagging Vcore might be a shorted CPU, but it sounds like that's not likely.

More realistically, I think I fried the VRM on the board. Any ideas which components on the board are most likely to have been ruined? Hopefully it's just the MOSFETs and not anything else. I can try to pull those and replace them at some point. My time with that board was short, but I liked it.

Is there any discoloration of the PCB? I once managed to "brown" a LGA 775 motherboard around the VRM area with some Core 2 Quad overclocking. That board had some very sturdy MOSFETs! Gigabyte Ultra Durable apparently means something.

Sometimes VRMs just die without too much fiery drama, it happend when I tried to overclock a Prescott with a MSI i875P motherboard. The CPU survived and the board would probably also work fine if I replace the VRM, I cant even tell which of the VRMs that is broken but thats an easy job for a DMM 😀

Is it worth to replace the vrm's ?isn't it too expensive ? I also have a board that has probably busted vrm''s and i was thinking about either replacing them,or trowing out the board and try and get a working 1.

I have a huge amount of failed motherboards from the same time period, VRMs isnt a problem. Bothering to do the soldering when I have other working Canterwood boards is 😁. My soldering skills isnt legendary so the board will probably look like shit when Im done, I will do it though sooner or later.

There's no discoloration, unless it's hidden under the backplate for the heatsink retainer.

By VRM are you referring to the MOSFETs or to the regulator chip? Once I get a decent soldering station I'll probably try pulling the legs up on the mosfets and hopefully find some sign of failure on them. Since it's not a dead short though I'm not sure if I'll find anything conclusive.
But if the regulator chip is bad, I don't have much confidence of being able to do anything with that.

shamino wrote:

There's no discoloration, unless it's hidden under the backplate for the heatsink retainer.

By VRM are you referring to the MOSFETs or to the regulator chip? Once I get a decent soldering station I'll probably try pulling the legs up on the mosfets and hopefully find some sign of failure on them. Since it's not a dead short though I'm not sure if I'll find anything conclusive.
But if the regulator chip is bad, I don't have much confidence of being able to do anything with that.

Its the MOSFET that fails in my experience, poke around with a DMM 😀