CPU Overheating Issue

[hungsolow]

Hi All,

I have a dual-CPU setup (KGPR-D16 MB, 2x AMD 6129 8 Cores). I have Thermaltake CLS0017 CPU coolers.

The tower I'm using is an ANTEC P193. Lots of fans, all of them running on lowest setting. I use the onboard video, the HD is in the lower cavity along with the power supply so I believe the air flow is rather good. I have three outtake fans (effectively surrounding the CPUs) and the intake is a massive fan on the side of the case. I keep everything rather clean and free of dust.

I use the system exclusively for scientific computing and I basically have all 16 cores working at 100% 24/7.

Here's the issue:

When I'm running the system at 100% capacity, I find the CPUs are at distinctly different temperatures. CPU0 (closest to the rear panel) hits around 65 degC, with the CPU fan having to rev-up constantly to maintain that temperature. CPU1 on the other hand is fairly cool at a mere 44 degC with the CPU fan never going above its lowest setting.

If I turn all fans to maximum (and make my office sound like an airport) I don't really see any difference in CPU0 temperature.

I really rely on this system for some time sensitive work (finishing my thesis), so I'm hesitant in both to fix or disturb things. I don't want my CPU0 to take down the system at some point, but I also don't want my system down for any period of time as I fix things.

(1) I was thinking about taking CPU0 out of its socket and checking the thermal paste (maybe the contact is poor).
(2) I was also thinking it might be a problem with the CPU itself (some defect).
(3) Maybe the placement of the fans or the location of the CPU on the motherboard is non-ideal.
(4) The CPU cooler being sub-par (anyone else have experience with this?)

I'm wondering what I should look for if I try to replace the thermal paste, how to remove the stuff and re-apply? Also, if the re-application of thermal paste doesn't solve the problem, where can I go from there?

The thing is I realize everything is okay as far as stability goes (no crashes) but it's disconcerting to see a 20 degree difference between two identical CPUs being loaded identically.

Any help and opinions would be greatly appreciate!

 

[o1die]

You have no problems. The cpu cores have different temps according to whether they're working. When one core is idle, it's possible the temp will vary be several degrees. Whether it's 44 or 65c, either temp is well within normal operating limits. I would leave it alone.

 

[hungsolow]

As I mentioned in the post, all of the cores are at 100%. Each CPU has identical load and thus should have roughly similar temperatures. I don't expect identical temps, but since both CPUs (and thus all 16 cores in the system) are running at 100% load, one shouldn't see a 20 degree difference between CPUs.

 

[o1die]

I don't see any info on your monitoring software. It's possible it's not accurate. I see alot of variation with coretemp with different boards. I don't use it with some boards as it differs too much from the bios reading.

 

[hungsolow]

I use about half a dozen different monitoring software (all agree within a degree) and I even measured with an IR gun. There's no doubt about the temperature. There is also no doubt about the CPU usage as I'm using multi-core software that definitely uses every bit of resource.

That fact is that one cpu IS running hotter under identical load conditions.

 

[Guest]

This issue isn't big.The operation tempt. of your cpu(44) is idol on load condition.Remember that temt. varies with core tempt. which is not permanent tempt.,at an instant of time it's low and at another its may be high so it's variable.As you mentioned work load on cores is 100% then obviously tempt. rises at high degree.
So not worry about it.If your work load is high then check out some liquid cooling.
Keep your cpu clean.

 

[kajabla]

also, 65 degrees is absolutely fine, not getting anywhere close to anything worth worrying about.

 

[Guest]

also, 65 degrees is absolutely fine, not getting anywhere close to anything worth worrying about.

If your tempt. rises gradually(65-70) and not comes down then its a serious problem.Then check out your cooling and fan configuration.Try to operate in open environment.And reboot at an interval of time to monitor the tempt.If nothing happen a tempt. remains constant then replacing your cpu is good option.