Temperature read-outs VERY 'jumpy'

IronHanded

IronHanded

IronHanded

Associate
Joined
30 Jun 2014
Posts
16
Hello guys.

I am running the following setup:
-4790K @ stock (4.4 turboboost)
-Dark Rock Pro 3 CPU Cooler
-Gigabyte Z97X Gaming 7

I have been stress testing the CPU with an array of tools (not using the new AVX2 instructions as they generate heat to Haswell that isn't relevant to everyday use.)

I have seen that my sensor readout for Core1 and Core3 are always 10C apart, Core1 will always be 9-11C (mostly 10C) hotter. I understand there is always a difference in core temp between individual cores but this seems to high to be normal.

It's very hard for me to judge 'how good' temps are for when I want to start overclocking, there's always this 10C difference in between cores. I don't really know what core 'to believe' to give a good readout. 70C or 80C? 80C or 90C? Too big of a difference to judge properly :(.

Does anyone have an idea about why this is or what can be done? I've never quite seen the difference be THIS big. Screenshot below of following situation:
-Asus RealBench v1.1 (latest, running for 30mins @ 22C ambient)

wrudr7C.png


measuring as you can see with latest HWmonitor + latest RealTemp.
Both tools display the dreaded 10C difference in between Core1/3.

Thanks in advance for any information regarding this situation, it's pretty damn weird to me!
 
Last edited:
Ryan-3 said:
Is your CPU cooler not fully down at one side? Uneven contact can cause this. :)
Click to expand...

I fitted outside of the case to ensure proper torque on all 4 screws, in any case all 4 sides are screwed equally tight.

Sadly its almost impossible to hold the hexagon nuts with a wrench in my case, space is too limited hence I had to take out mobo entirely.
 
Just focus on the hottest core, the rest of them are irrelevant...

I believe the cores run at significantly different temperatures (**** TIM/manufacturing/sensor variation aside) due to location on the die. Core #1 and #2 are sandwiched in the middle of two other cores and cache so they will run the hottest, Core #0 and #3 have the onboard GPU or system agent etc to the side of them which will run cooler and offer room for the heat to spread. I'd imagine if you used the onboard GPU and ran a stressful game at the same time as Prime95 one of those lower temp cores might run hotter, or if you add extra voltage to GPU/system agent etc

Die pic here:
http://wccftech.com/review/intel-haswell-core-i7-4770k-review-dz87kl75k-motherboard/
 
Last edited:
IronHanded said:
I have seen that my sensor readout for Core1 and Core3 are always 10C apart, Core1 will always be 9-11C (mostly 10C) hotter. I understand there is always a difference in core temp between individual cores but this seems to high to be normal.
Click to expand...

I've had issues with an Ivy that was showing 1 core between 20-30 degrees cooler than the other 3 under heavy load (it was core 0 that was the cool core). It's still with Intel, being looked at, but they suspect its just a bad IHS contact. Some difference between the cores is expected and if your temps are not getting into the danger zones, I wouldn't be too concerned.
 
setter said:
Tbh every ivy/haswell chip has bad contact due to intels cheap paste job.
Click to expand...

This isn't the first I read about that. Rather sad that the quality of their TIM seems to decrease instead of gaining quality.

Too bad that I'll have to look at the 'highest core temp', it means my temperatures are far from good at stock, not a lot of headroom for any OC'ing. :(
 
Core 0 on my 4670k is 10 degrees lower than all the others as well. Bloody annoying but apart from a de-lid there's not a lot you can do.
 
Thanks for the replies guys.
Before I will let this thread die I'd like to ask one more question.

Are my temps good in your opinions? Basing on running stock speeds with the Dark Rock Pro 3, MX-4 paste I can't help but think they are too high.
Already breaking 70C with Asus RealBench which isn't really high in load. OCCT/Prime95 goes through the roof almost instantly, 85-90+ within 1 minute...
This doesn't seem right even with AVX2 when using a 250W TDP rated cooler?

I'm not really sure as to what temps I should expect with this combo, anyone has 'expected readings' of a 4790K @ stock with enthusiast level air cooler such as the dark rock pro 3?

Again thanks for the info, it atleast made my mind more at ease, I think most of you can understand worrying about this stuff out of proportion. :rolleyes:
 
Last edited:
That does seem pretty damn high for a good air cooler on a 4790K which afaik is designed to be able to be used with the shoddy stock intel heatsink. Have you tried remounting your heatsink?

And again the temp differences between cores is fairly normal, my 2550K's core 0 is 10C+ cooler than all the other cores despite being Sandy bridge which is soldered together rather than using TIM.
 
Also depending on system setup its not unusual with 4+ cores setups that one core will hardly get used at all generally when not benchmarking or running intensive software and another core will often be the one that general single threaded stuff, windows background processes, etc. gets dumped on so its not unusual to see 2 cores that are quite a way apart even with the system mostly idle.
 
Disco_P said:
That does seem pretty damn high for a good air cooler on a 4790K which afaik is designed to be able to be used with the shoddy stock intel heatsink. Have you tried remounting your heatsink?

And again the temp differences between cores is fairly normal, my 2550K's core 0 is 10C+ cooler than all the other cores despite being Sandy bridge which is soldered together rather than using TIM.
Click to expand...

I can try remounting it. Problem is that I don't have many references as to what kind of performance I should expect with a 4790K.

People seem to have mixed experiences wether the 4790 runs hotter/cooler than the 4770... it may be very chip dependent I guess?
 
You must log in or register to reply here.
Back
Top Bottom