15C difference between CPU cores while stress testing, why?

[bubasvabac]

I am experiencing a huge differences in core temperatures, delta between core #1 and #4 averages is around 15C while maximum difference is 20C when stressing the CPU with Aida64.
I am using an i5 4690K overclocked at 4.5GHz with a 240mm AIO.
Does anyone know why is this happening, is it because of a bad TIM application or could it be something else?
Btw i tried reapplying thermal paste and reseating the cooler but the temps stayed the same.
Pics:
http://
http://

 

[Barty1884]

Are those sustained temps? Or just rare spikes?

If spikes, I wouldn't worry about it. The screenshot appears to have been started during a stress test? Since minimums are ~60'C, it's hard to judge.

If these are sustained temp variances, it's definitely above the norm (a couple of degrees, maybe as high as 8-10'C between hottest/coolest core wouldn't be unheard of). If you're sure you've applied thermal paste & seated the cooler corrected (and not tighter on one side than the other), then chances are good this is an issue with the stock TIM between the die & IHS.

While the stock TIM isn't great in general, as high as 15-20'C variances between cores is pretty extreme.

 

[CompuTronix]

bubasvabac,

On behalf of Tom's Moderator Team, welcome aboard!

Intel’s specification for Digital Thermal Sensor (DTS) accuracy is +/- 5°C. This means deviations between the highest and lowest Cores can be up to 10°C. Deviations on processors that have an uneven application of TIM might exceed 10°C by several degrees.

However, as Barty1884 pointed out, if your cooler is properly installed and the TIM is applied correctly, then the problem is under the IHS. Keep in mind that your Devil's Canyon 4690K was launched in the 2nd Quarter of 2013, so yours could now be over 5 years old. Intel reportedly changed to a "polymer" TIM after the temperature disasters of the Ivy Bridge and Haswell processors, so Devil's Canyon is now mature enough for evidence of TIM failures to become apparent.

I very recently delidded a 4690K with exactly the same thermal behaviors you've described. Before delidding, deviations between Cores was about 18 to 20°C. After delidding, deviations under the same test conditions decreased to 6°C, which is well within the 10°C (+ / - 5°C) spec. Overall Core temperatures also decreased by about 16°C, so it was a win-win scenario.

You might want to consider delidding, which will fix the problem. If you've never delidded and don't want to try the razor blade method, you can safely delid with a "delidding tool" such as the Rockit 88 - https://rockitcool.myshopify.com/

Delidding requires that you use only liquid metal TIM between the Die and IHS. Typical silicon TIM will fail in a relatively brief period of time. A process known as “pump-out” will cause silicon TIM to ooze out from between the Die and IHS due to thermal cycling. A highly recommended liquid metal TIM is Thermal Grizzly Conductonaut - http://www.thermal-grizzly.com/en/products/26-conductonaut-en

Silicon Lottery - https://siliconlottery.com/collections/all/products/delid - is a company that tests, bins and sells overclocked, delidded "K" CPU's. They also offer professional delidding services, which is always a good option.

Also, we have a Sticky you might want to read: Intel CPU Temperature Guide - http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html

Once again, welcome aboard!

CT

 

[bubasvabac]

First of all thank you very much for answering my question,
I came to the same conclusion myself as both of you pointed out that the problem is with the TIM between the IHS and the die.
I already ordered a cheap deliding tool and an IC Diamond thermal paste.
I dont want to frequently reapply liquid metal, also i heard that liquid metal reacts and damages the die with time so i opted out for a thermal paste.
IC advertises Diamond as a non silicon paste that doesnt “pump-out” with time so i hope i made a right decision.