I think my X2 4800+ Brisbane is dying....

[Nixxon]

Hello. I am (was?) the happy owner of a 65nm Athlon X2 4800+ (Brisbane core) on a KN9 Abit motherboard... A month ago my heatsink (Arctic Freezer 64 PRO PWM) fell onto the video card after one of the lugs from the retainer broke. The computer was on, I wasn't at home, my girlfriend phoned me telling me there was a strange noise from the computer, and now there's an alarm ringing... I immediately knew what had happened, and so she cut the power from the system quickly...

I feared the worst, but luckily the video card was not damaged and - after repairing the damaged AM2 retainer and fitting the old stock cooler - the CPU seemed well. I only tested a few minutes of SuperPi, it ran without errors and that was it. After a long search I finally found yesterday a Thermalright retainer (AM2RM rev.B), and was able to put back the Freezer... I also got a syringe of Arctic Silver 5, and so I wanted to check the temps... But running two instances of Prime95 (one for each core) gets the CPU temperature to 90 degrees C fairly quickly and the computer shuts down!

I have tried lots of things... Updated the BIOS from ver. 12 to 14 (temperature problems were among the fixes on the abit website)... Tried different programs to read the temperature... CoreTemp, SpeedFan, Sandra, AbitEQ...
I also put back the AMD stock cooler... the same. The computer is OK for light load, also some gaming... But two instances of Prime95 or SuperPi are getting the overheat protection to kick in after a few minutes only... And I could do that for days before the heatsink incident...

The CPU doesn't seem that hot when I touch it. The Freezer lets quite a bit of the heatspreader uncovered. And I know how 90 degrees C should feel. I wonder if the thermal diode was damaged when the heatsink fell off... Or maybe I crushed the processor when I attached the Freezer (is that even possible?)...

Some of you may say that incorrect temperature readings are a common problem with Brisbane cores. But they were just perfect in the beginning.
So, that's my story. I'm afraid it's a bit too long, maybe boring, and mildly incoherent in some places. But I've been trying to solve this all day, it's 3:02 AM where I live, and I can't shake the feeling that my CPU went south... Any advice will be greatly appreciated. Thanks in advance!

 

[muddocktor]

It sounds to me like that new HRM from Thermalright isn't holding the heatsink flat against the processor for some reason. How were your temps when you had the stock heatsink installed with the original (and broken) motherboard HRM? That's hat I would try if you are having problems with the TR HRM possibly. Try installing the old stock HRM and the stock heatsink and see if your temps climb out of sight like you are seeing now when you Prime or Orthos it.

 

[Nixxon]

Thanks for your suggestions.
First of all, I didn't measure the temp after installing the stock HSF w/ the broken retainer... I did not have fresh thermal paste and so I installed it with the old paste that was on the heatspreader and HSF ... So I didn't put much stress on the CPU because of that, I only ran a few minutes of SuperPi (a single instance) to quick-check if the processor can still do correct calculatios (and it did).

The Thermalright HRM not holding the heatsink flat against the CPU is a possibility with the stock, but the Freezer has a base that's a bit smaller than the heatspreader... And I clearly saw the Arctic Silver being squished towards the edges when I closed the heatsink lever. And it takes considerable effort too, so I'm sure it's not from loose contact between HSF and CPU.

I'll try the broken retainer too, I'll try everything before buying a new CPU... I'll even run a test with the mainboard out of the case... 90 degrees C on the CPU die should make the back of the mobo pretty hot, I guess...

So far I think there are 2 possibilities: the processor is reporting wrong temperatures (maybe its sensors were damaged when it worked without cooling) or there's some damage to the interface material between the silicon and the heatspreader, and the CPU die is really going over 90 degrees while the heatsink (and heatspreader) remain barely warm...

 

[Nixxon]

Thanks muddocktor - your advice got me on the path to solve this.
Everything is OK now with my CPU temps... But I haven't got the slightest idea why it is all working now...

I disassembled my computer... Checked if everything about the motherboard and CPU is ok visually... Then tested again with the old (broken) retainer and the stock HSF (even using the old Arctic Silver that was smeared on them)... And the temps were OK...
EVRIKA I said to myself... The retainer from Thermalright must be the culprit. But to make sure of that I mounted it again... And there it was, working perfectly even with the "new" retainer... As I said I have no idea why it is working now... It seems perfectly symmetrical, so I don't know if I mounted it the same orientation as it was before... Anyway I'm not going to reverse it again just to check... I'm happy that it's all going as it should.

I then tested with the Freezer, but I got mixed results... The temps were as high as with the stock HSF (around 63 degrees C in full load), it was much louder at full speed and when I poked around with my fingers to "feel" the temp of the heatspreader the temp jumped to about 75 degrees (I haven't tilted the Freezer, just touched the IHS)...

So I'm using the stock cooler now, very happily Prime95-ing right now at 65 degrees (my room is hotter than it was yesterday). And I can say that the Freezer is one of the worst computer parts that I bought... Breaking the retainer and causing all these problems really offset any of the advantages it might have had in the beginning...