6950x Cores 2 and 3 running 4c hotter than the others. What's normal?

[ianross]

Hey all,

I'm a bit of a perfectionist and I'm wondering why two cores on my 6950x are running hotter than the others by about 4c. I'm running max turbo boost (no OC) and my idles are around 34c on all cores except 2 and 3 run at around 38c.

I have no intention of lapping or deliding, but I am wondering if the hotter cores are due to either air bubbles or lint getting in the way. I don't want to reseat and grease my Noctua NH-D15s just to find out that cores 2-3 are meant to run hotter.

I'm viewing my temps with HWmonitor and test stressing with CPU-Z. The differences are consistent at stress levels.

Could anyone provide some info insight into this, especially 6950x users?

 

[Janus67]

3-4c is a perfectly fine amount of difference between cores. if they were 10-15+ different then remounting or something may be recommended. Not all cores run at the same temps.

 

[ianross]

3-4c is a perfectly fine amount of difference between cores. if they were 10-15+ different then remounting or something may be recommended. Not all cores run at the same temps.

Hey thanks. I know it's perfectly fine and I don't need to worry but I am a perfectionist.

I'm trying to find out if the 6950x runs cores 2 and 3 hotter by design or if it's my heatsink contact?

Anyone else with a 6950x getting the same temps across all cores or are 2 and 3 always running hotter?

 

[Janus67]

every CPU will be different, some will be off by 2-3c, some by 4-5c and on different cores. Intel didn't make the 6950x to have the core 2 and core 3 run hotter, it's complete luck of the draw and to be honest you did very well in the silicon lottery to have only two cores out of 10 be off by only a couple degrees C.

 

[EarthDog]

My 6950x is a few C apart between the cores as well. Perfectly normal, like Janus mentioned.

 

[Woomack]

3-4*C is actually pretty good. Most chips have bigger differences between cores.

 

[FTC]

Note that many things can affect this difference, beginning with the calibration of each individual core sensor (which can be a bit off for some cores), the quality of the contacts die-ihs-cooler right above the core, the instructions being processed by each individual core, the geometry of the cooler and the cooler base vs where the contact is better (usually better right under the heatpipes that are better refrigerated), also affects what is near a specific core ... i.e if a core is just near the IMC and the process being carried on does not use much the memory subsystem, that core may tend to be cooler than another that is nearby to another core being loaded... and there are also some other intrinsic to each core factors (like the silicon lottery), i.e doping could be better deposited on a given spot of a die, creating a cooler / hotter core.

So in summary, it does not make sense to pursue differences smaller than 10ºC between cores. As said in previous posts, if you find greater differences then a cooler reseat may be something to try...