Core i5 2500k Temperature Difference Amongst Cores

[handsome_dave]

Can someone please tell me if a 10-12 degrees Celsius temperature difference is normal between cores? I'm currently at

Core 0 - 30C
Core 1 - 24C
Core 2 - 33C
Core 3 - 21C

Is this Normal? If I overclock, I think that one core would also sky rocket in temperature:S

 

[Soundmanred]

As many other threads have said, yes its normal.
If there is a large discrepancy in range, you may want to reapply your TIM and reseat your heatsink.

 

[tweakboy]

The reason the temps are different is because the CPU or Heatsink is not flat and is a bit concave. I have a 8c difference between cores this is why people LAP their CPU's You take grit of sand paper and sand away your CPU until its smooth and flat, then that temp difference will be closer about the same for all cores.

 

[RobDickinson]

I'm going for the fact some cores arnt working as hard as others.

Hit it with some threaded burn in software see how the temps scale.

Mine currently are (at pretty much idle)
22
17
22
20

 

[Kenmitch]

Stock cooler? If so did you double check the push pins to make sure they are seated correctly. Is this new fresh build? TIM may need to break in if so.

 

[handsome_dave]

I've built this about 3 months ago, using Artic Silver 5 and corsair A50 heat sink. I've reseated it about 3 times with the same results. Core 2 consistently shows 10-12C higher then core 0. Should I be concerned this may cause long term damage?

 

[RobDickinson]

ffs intel cpu's can take near 100c dont worry about it.

 

[Diogenes2]

As previously stated, as much as 10c difference between cores is reported quite often ..
It appears to be normal for SB ..

 

[4ghz]

As previously stated, as much as 10c difference between cores is reported quite often ..
It appears to be normal for SB ..

I wouldn't even limit it to sandy bridge. I have an i3 530 where both cores idle within 3c to 4c of each other. I have an i3 540 where the idle difference is 12c. But under load the gap closes to between 7c or 8c. Just seems to be luck of the draw.

 

[TurtleCrusher]

If it means anything, my core2quad Q8200 had core #2 nearly 9c hotter at idle, 4c hotter at load. I reseated the arctic freezer pro v2 so many times with different TIMs. Turns out its just the sensors...

Dont worry if you tried and tried. Its probably the sensors.

 

[BonzaiDuck]

The core "TJunction" sensors were never meant to be spot-on accurate. So you can expect to see variation for that reason, even as you would expect variation for the cores being loaded differently.

Even so, if three out of four cores cluster around their own average value at both idle and load, it is possible for one core to be out of whack.

And as for idle temperatures, Intel had issued a disclaimer a few years back that the sensors were not meant to report idle values accurately in any case. This held them harmless from the criticism that some sensors were "stuck" at values higher than they should be, rising only above those stuck values as actual temperatures exceeded them.

 

[amenx]

Some good info by RealTemp author on Intel temp sensors and how and why they may vary:

http://www.techpowerup.com/realtemp/docs.php

 

[Puppies04]

I've built this about 3 months ago, using Artic Silver 5 and corsair A50 heat sink. I've reseated it about 3 times with the same results. Core 2 consistently shows 10-12C higher then core 0. Should I be concerned this may cause long term damage?

C2 is the hottest of my cores aswell, reseating/reapplying tim doesn't help.

 

[handsome_dave]

Thanks for everyone's helpful response and clarifying that I have nothing to worry about I guess my paranoia got the best of me.

 

[Idontcare]

I've built this about 3 months ago, using Artic Silver 5 and corsair A50 heat sink. I've reseated it about 3 times with the same results. Core 2 consistently shows 10-12C higher then core 0. Should I be concerned this may cause long term damage?

I've lapped both my 2600K and H100, reseated multiple times, using NH-T1 TIM, and my core 0 still reads 6-8C lower than the other three cores.

Whatever is going on, it appears to be under the IHS surface, not external to it.

 

[Kenmitch]

My core 0 has been as low as 8*C which I'd never believe

As others have stated idle temps aren't accurate at all. I once had a e5200 that wouldn't go below 35*C at idle no matter what.

My 2500k's core 1 & 2 run about 5-6*C warmer under load. At idle they're pretty much all the same tho.

I'm wondering if the internal things such as mem controller in the SB's heat up certain cores more. Kinda also wonder if using the onboard video would also influence the core temps.

 

[Idontcare]

I'm wondering if the internal things such as mem controller in the SB's heat up certain cores more. Kinda also wonder if using the onboard video would also influence the core temps.

That's a good point, if Core0 is physically the core next to the IGP and that IGP is disabled then we would expect Core0 to be the coolest of all the cores simply because it has a much larger effective surface to dissipate heat through.

Come to think of it, my hottest core is consistently Core2, which is also easily explained based on the thermal restrictions imposed by Core1 and Core3 (core3 being next to the PCU which is also likely to be cooler like the IGP).

 

[tweakboy]

As I said earlier, its concave. You have to LAP the CPU to get even temps.

Thank you and gb

 

[Mallibu]

As I said earlier, its concave. You have to LAP the CPU to get even temps.

Thank you and gb

And as people said they already done it and still the 'issue' remains.
So i suggest first to learn how to read and then adopt the "Thank you and gb" attitude.

 

[tweakboy]

And as people said they already done it and still the 'issue' remains.
So i suggest first to learn how to read and then adopt the "Thank you and gb" attitude.

Well then the heatsink is concave. Its not getting full strong contact with CPU. Its normal forget about it and use your PC.

Thanks malibu beach. zuma

 

[IntelEnthusiast]

On the 2nd generation Intel® Core™ Processors the Thermtrip# (point at which the processor get too hot and shuts down) is 130C. In other words you are going to have to really push it to get to the point where the CPU will shut down. Something else most likely will shut down long before this. http://www.intel.com/content/www/us/en/processors/core/2nd-gen-core-desktop-vol-1-datasheet.html p.65

Christian Wood
Intel Enthusiast Team

 

[Idontcare]

As I said earlier, its concave. You have to LAP the CPU to get even temps.

Thank you and gb

Because the SB die is not symmetric, we would not expect the thermal dissipation itself to be symmetric, which means we ought to expect a temperature disparity amongst the cores even if the surfaces are perfect.

If the IGP is not in use, the core closest to the IGP ought to be the coolest, followed by the core on the far right which is closest to the PCU.

The 2nd core from the left should be cooler than the 2nd core from the right because the core farthest on the left is cooler than the core farthest on the right.

The core 2nd from the right ought to be the hottest, all things being equal. (assuming process-induced variation within-die is minimal)

 

[darkpaul13]

i have a problem with my i5 2500k .Core 3 seems to have a problem.

Core 1-36
Core 2-34
Core 3-59
Core 4-37

 

[Idontcare]

i have a problem with my i5 2500k .Core 3 seems to have a problem.

Core 1-36
Core 2-34
Core 3-59
Core 4-37

Welcome to the forums darkpaul13:thumbsup:

Core 3 may have a poor thermal sensor. What matters is if the discrepancy goes away as the core heats up and approaches TJmax.

What do your temperatures look like when running LinX? (see the OC sticky in this subforum for a download link)