different core temps bad?

[Zodan]

I just built a system around the i5-2500k. I'm using a cooler master 212 plus heatsink and I've been watching the temperatures of my cores using both core temp and real temp. Haven't started overclocking yet btw.

I'm seeing on average:

idle: 28, 31, 32, 35
load (12 hours of prime 95 small test): 47,52,53,53

I think in general these temps are pretty good but it bothers me that core #4 is so much higher than the others at idle...although it seems to even out under load.

Is this a concern? Are differences in the core temps normal or did I do something wrong when installing my cpu/heatsink?

thanks!

 

[d4a2n0k]

Totally normal, nothing to worry about.

 

[bryanW1995]

I would love to have only a 6c delta between my hottest/coolest cores. I used to try endless reseats, but unless one of the cores is dangerously high nowadays I just leave them alone.

 

[Drifto570]

Have no fear
normal temps =)

 

[BonzaiDuck]

We've been through this in other posts. I mentioned it extensively in a thread in this particular forum -- posts as recent as today or yesterday -- with a title about "Stress-testing, Windows 7 and BSODs."

First: A known assumption that is certainly correct: These sensors are not accurate. Maybe, they vary from the real temperatures by +/- 4C degrees. Even if their accuracy falls within that range, it is reasonably likely you will have one core with a sensor that reports outside the range, or different than the cluster of the other three core temperatures.

Second: as much as Intel hinges its TM1 thermal throttling to the 72.6C spec associated with "TCASE," we haven't found any monitoring software which reports "TCASE."

Third: There is a rule-of-thumb propagated by a Tom's Hardware article -- very authoritative -- which noted for the LGA-775 C2D and C2Q processors that "TCASE" would be less than an average of the core temperatures by about 10C degrees.

We do not know with any certainty whether this same spread or difference holds true for the Sandy Bridge cores.

So like the last poster said about the outlier core temperature different from a cluster of the remaining three -- have no fear.

With regard to over-clocking, with the information we have available, you might want to err on the side of caution as temperatures that you can "see" in monitoring software come close to -- or just above -- the 72.6C spec.

 

[amenx]

The temps are supposedly accurate higher up the scale, as they approach TJMax, because for CPU makers like Intel, thats all that matters. They couldnt give a hoot about enthusiasts fidgeting with idle temps. If the CPU runs and its overheat protection functions, nothing else matters to them.