Do CPUs just go bad and overheat?

[Craig234]

8 year old AMD 965 II - started overheating, I figured it's way past the thermal paste going bad and re-applied that. But - it just overheated again on an MMO, which is when it was doing it before.

It's still very dusty in the case, so I could look at that.

But could it be just the CPU has gone bad?

 

[Doom2pro]

I had a GTX 570 do that, I delidded it and applied fresh compound but it didn't help much, and about a month later it died.

 

[XavierMace]

Generally if it's overheating it's either time to clean out the case or reapply paste, or both.

 

[MajinCry]

Do the Phenom II CPUs have solder or thermal paste between the silicon and that metalish top? Could be that the goop's gone bad.

 

[bigboxes]

I don't think I've ever killed a CPU. Usually it's other components that died and the platform became too old to keep investing more into it. I keep an Athlon X2 on top of pad of post-it notes.

 

[Lorne]

I have an Ph2 550 that started running really hot.
I assumed that maybe the disabled 2 cores were acting up.

 

[Crono]

The fans (including heatsink fan) and dust are the first culprit I'd rule out before reapplying paste, though that would be what I would rule out next.

 

[VirtualLarry]

I've run across older OEM AMD AM2/AM2+ rigs, usually with a 5000+ or 5200+ CPU in them, that run inexplicably hot. They seem to just "get old" and start running hot. Even with a fresh paste application, and a brand-new retail 125W AMD 4-heatpipe heatsink.

It could be the TIM between the die and the heatspreader drying out, too. Maybe a de-lid is in order?

 

[lxskllr]

I had the heatpipes go bad on a heatsink once. It was on a laptop, so I took off the keyboard, and thermal epoxied a different heatsink to the chip. It worked til it was fully decommissioned.

 

[lopri]

More likely the motherboard's VRM is giving up. Assuming it has worked for 8 years, it is perhaps time to let it retire without asking questions.

 

[UsandThem]

They do when your heatsink is clogged up with dust. Clean out the dust and you will know for sure.

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/heatsink-removal.2498452/#post-38719468

 

[Craig234]

I've run across older OEM AMD AM2/AM2+ rigs, usually with a 5000+ or 5200+ CPU in them, that run inexplicably hot. They seem to just "get old" and start running hot. Even with a fresh paste application, and a brand-new retail 125W AMD 4-heatpipe heatsink.

It could be the TIM between the die and the heatspreader drying out, too. Maybe a de-lid is in order?

I have no idea what you said there in the last part.

 

[Crono]

I have no idea what you said there in the last part.

For most modern desktop CPUs the actual CPU (the die) is enclosed within a metal enclosure, the IHS (Integrated Heat Spreader). In effect, it's a heatsink before the heatsink. The die can be soldered to the IHS or have TIM (Thermal Interface Material) to transfer the heat from the die to the IHS.

I've never delidded (removing the "lid" or IHS, usually to apply better TIM or replace old TIM) a CPU myself, though. I like to think of it as CPU trepanation.

 

[Craig234]

They do when your heatsink is clogged up with dust. Clean out the dust and you will know for sure.

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/heatsink-removal.2498452/#post-38719468

This is dust like you haven't seen. So, maybe.

Now, picture that case cover off, which it was for years until today.

 

[UsandThem]

This is dust like you haven't seen. So, maybe.

Now, picture that case cover off, which it was for years until today.

Man, that is craziness right there. Unplug that PC, take it outside, and hit it with a compressor or a shop vac. I'm just amazed your PC has lasted that long with all that dust that has to be clogging your PSU and CPU heatsink.

 

[BeepBeep2]

No, that doesn't happen.
And all Phenom II CPUs are soldered under the IHS - do not suggest you try to delid them.

Check case airflow and CPU fan RPM - what are your temps like? HWMonitor is a great program for checking these things.

 

[Craig234]

No, that doesn't happen.
And all Phenom II CPUs are soldered under the IHS - do not suggest you try to delid them.

Check case airflow and CPU fan RPM - what are your temps like? HWMonitor is a great program for checking these things.

It shut down about 117c today running the MMO. After that, it's showing min 49c max 100c.

 

[VirtualLarry]

No, that doesn't happen.

I'm not so sure. After a number of years of solid usage, the dielectric layers of a CPU can break down slightly, resulting in greater leakage currents, which results in greater power usage, and thus greater temps.

I believe that I've observed that, on 5+ year old OEM boxes with AMD CPUs.

 

[lopri]

117C..?! I did not know that CPU is capable of logging that kind of temp. That is not even "Core Temp"..

 

[Craig234]

117C..?! I did not know that CPU is capable of logging that kind of temp. That is not even "Core Temp"..

It's hit about 121c before shutting down. I thought it was the paste. It only happens running MMO's.

 

[Crono]

Don't forget to wear a dust mask when cleaning the case out.

Took me over an hour to dissasmble and dust (using a compressor) to my satisfaction a computer yesterday that I had built for someone else. It's just a year old, but with high airflow it had a good amount of dust in it. Worth it, though, especially since it has two R9 390s and a 1000W PSU I'm going to use in another build.

Best not let your computer go a few months without dusting, though, and if you can you should build a PC with dust-proofing (positive pressure fan setup, easily removable magnetic dust filters, easily removable side panel, good cable management to make for better access, etc.) in mind.

 

[mikeymikec]

I've only ever seen one CPU die (ie. it was unquestionably that), and I suspect it was because the user OC'd it. K6-2. My old P3-866 was in use until 2013 in a 24/7 server (it wasn't 24/7 from new, I pulled it from a customer's computer when they had finished with it in 2005); I only decommissioned it because I felt it was time.

OP, has this CPU been OC'd? ever?

 

[pcgeek11]

Usually not the CPU unless it has been overheated too much. I started with the old 8088 CPUs and have never had one die on me. I sold the old 8088 say 2005 and it was still working fine.

 

[Craig234]

I've only ever seen one CPU die (ie. it was unquestionably that), and I suspect it was because the user OC'd it. K6-2. My old P3-866 was in use until 2013 in a 24/7 server (it wasn't 24/7 from new, I pulled it from a customer's computer when they had finished with it in 2005); I only decommissioned it because I felt it was time.

OP, has this CPU been OC'd? ever?

No. But 8 years old.

 

[mikeymikec]

No. But 8 years old.

I wonder whether pcgeek11 is correct in your case. An 8 years isn't old for a processor in my experience.

Do the CPU temps in the BIOS concur with whatever software you're using?

The other thing that I find rather suspicious is that I'm pretty sure thermal shutdown should have occurred at 90C or thereabouts.