Uneven temps

keller772 · November 6, 2014

So I'm working on dialing in an overclock on my 5820K and I was noticing that the temps of my cores seemed to vary quite a bit. So much so that I thought I should confide in the knowledge of the good people on this forum! Anyways I'm running a 5820K currently at 4.2 GHz and 1.29v, which to me seems high for only 4.2GHz, on a Gigabyte X99 gaming 5 (really sexy board in my opinion). I was running Prime95 while monitoring the temps with RealtempGT. I then noticed that two of my cores were significantly hotter than my two coolest cores. The hot cores were at 90 and 89 (1 and 3 on realtemp) and the cooler ones were at 79 and 78 (2 and 6). I don't know if the order that realtemp gives the readouts matters I'm just assuming that it does. So I was wondering why I might be seeing this gap in temperatures within the CPU? I'm assuming it's a thermal compound issue but I'd be open to some other opinions, which is why I am here. Just in case you'd like the other two temps they were 83 and 86 on cores 4 and 5 respectively. So should I reapply the compound in a specific way? Or is this something else entirely? All in all the system has been working perfectly fine, just want to be able to keep it running fine.

Faceman · November 6, 2014

Was your voltage on adaptive?

Temperature disparity between cores is common. 11C is on the high end of the spectrum, but it is still within the realm of possibility. Try setting your voltage to manual/constant/override/static if you haven't already. That should close the gap and give you more consistent numbers.

Always a good idea to set voltage to manual before any kind of stress test.

keller772 · November 6, 2014

I'm fairly certain I had it set to manual or whatever the non adaptive one is. But you're making me doubt myself now that you mentioned it! I'll have to check that out next time I reboot.

I was also wondering if a certain method to the thermal paste might have an effect. All I did was a pea shaped dot in the center of the heat spreader. But since this is 6 cores do you think a line might make a difference? Something to follow the die more closely to ensure good covering with the paste?

SuperPug · November 6, 2014

I'm fairly certain I had it set to manual or whatever the non adaptive one is. But you're making me doubt myself now that you mentioned it! I'll have to check that out next time I reboot.

I was also wondering if a certain method to the thermal paste might have an effect. All I did was a pea shaped dot in the center of the heat spreader. But since this is 6 cores do you think a line might make a difference? Something to follow the die more closely to ensure good covering with the paste?

You can try reseating the Cooler and reapplying thermal paste but i dont think that's causing the problem. If anything it could be a bad contact between the cpu heatspreader and the actual core.

BTW prime95 is an unrealistic stresstest in terms of real world application so i'd test rather by running different games and maybe cinebench or something like that. I bet the temps wont spike that high then...

typographie · November 6, 2014

I was also wondering if a certain method to the thermal paste might have an effect. All I did was a pea shaped dot in the center of the heat spreader. But since this is 6 cores do you think a line might make a difference? Something to follow the die more closely to ensure good covering with the paste?

I'm not sure how much the method itself might affect this, but I think uneven application in general could. The "pea" method is basically what I'd use, though a pea might be a bit too much. The size of a grain of rice might be a bit better way to visualize it.

I, too, get fairly uneven temperature distribution between the cores on my system, but I don't think the delta in my case is ever as large as what you're describing.