My computer mysteriously shut itself down with no warning a few times over the past 24 hours, both at times when it was not being used but left on (incl. overnight) and a couple times when playing a game, so under at least moderate load.

This led to me installing a temp gauge (Realtemp 3.7) and a stress tester (prime95). The first run of the stress test showed a between-test phase idle temp of 79C but quickly pushed the cores to 105C at which point I stopped the test.

Then I opened up the tower and did a pretty thorough compressed air cleaning and made sure all the fans were spinning.

Oddly, all subsequent tests I ran didn't push the temp beyond 80C, and averaged around 40C. And for the next few hours, temperatures were just fine, including during gaming. Then just minutes after I left the computer alone, the fans kicked into high gear and when I checked the temp it was climbing over 90. I've copied a couple pieces of the log to illustrate (the temp's are in C; the far right column is CPU load %.

Here, while gaming:

19:56:15 3372.37 36 30 40 36 46.7

19:56:30 3342.44 36 33 40 39 53.1

19:56:45 3342.43 36 31 43 35 54.0

19:57:00 3332.46 37 34 41 39 60.2

And then later, just after stopping using the computer:

22:57:15 1596.39 87 86 87 89 3.2

22:57:30 1596.39 92 92 92 92 3.6

22:57:45 1596.39 88 88 89 89 3.4

22:58:00 1596.39 95 95 95 95 3.9

It seems like there's an obvious discongruity between load % and the CPU core temps. Could my cooling system just be sporadically failing?

Basic specs: Intel Core i5 3450 Quad 3.1 GHZ, 8gb DDR3, AMD Radeon HD 6850, CoolIT ECO A.L.C water cooling.


  • reseat everything because those temperatures are way to high if you water cooled correctly. Your CUP could almost BOIL water...You have major problems... – Ramhound May 24 '13 at 0:48
  • Any idea why the heat spikes are sporadic and unrelated to the core load? – Eric May 24 '13 at 3:00
  • When the temps get that high, doesn't the CPU get throttled back to help reduce temps. Then when that does not work and temps do not go down, thus mysterious shut down at low load seems not so mysterious. – Carl B May 24 '13 at 6:41
  • Did you add thermal paste during installation? Just curious as this unit ships with paste(or pad) preapplied. Too much is just as bad as too little. – Carl B May 24 '13 at 6:52
  • @Eric - It could be one of any number of reasons. Have you done what I suggested? Your system should not even get close to those numbers. I assume you have not overclocked your CPU since you didn't mention this fact. – Ramhound May 24 '13 at 11:29

It sounds like your CPU heatsink / block is not properly attached or you are not using enough thermal paste.

Does RealTemp report LOG in the Thermal Status area? That means your CPU has logged at least one thermal throttling episode. If it shows HOT, that means thermal throttling is fully engaged.

How much thermal paste did you use and how did you apply it. You need to do this again.

  • The last part of the question seems to indicate water cooling - apparently a self contained model – Journeyman Geek May 24 '13 at 15:01
  • This heatsync came pre-installed on an assembled system, so I assume they just used the pre-applied paste. Thankfully it's still under warranty for a couple months, and after contact the seller last night with the details, they quickly offered to replace the unit. – Eric May 24 '13 at 18:48
  • Just out of curiosity, since I'll still have the failed unit after uninstalling it, is there any visual clues one can use to see if a water-cooling unit has failed? (other than leaking, I guess, of which there was none) – Eric May 24 '13 at 18:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.