Overclock.net banner
1 - 6 of 6 Posts

·
Registered
Joined
·
106 Posts
Discussion Starter · #1 ·
So recently I've been concerned by my temperature variation between cores, I can't remember but i'm pretty sure when i first had this chip almost a year ago now it wasn't doing this, recently i reformatted and decided to bench a bit again and noticed this. I saw this initally and restated the Cooler with no effect on the temperatures so i decided to test further

My methods are far from perfect or statically relevant (yet), but this is just my first test i plan on running a few more then resetting my headsink again to rule out my thermal paste or application method

Currently i ran using intel burn test v2.54( default test (10 with 1024mb ram)), running with realtemp recording every 10 seconds, then just basic calculations were preformed.

Currently I'm running a i5-3750k at stock voltage with 42 time multipliers, running a push H100 Setup with stock fans and AS5 thermal paste

But it appears from my inital test that my core 0 is running almost 10 degrees cooler then core 1, and 5 degrees cooler then 2 and 7 cooler then 3.

On my second gen i5 2500k I don't see this kind of variation between cores, even at idel i see 18-20 on core 0 and 28-32 on core 1.

Count is 194 readings at 10 second interval (test time was 3.2 mins)

Calculation Core 0 Core 1 Core 2 Core 3
average 53.89 62.63 57.62 59.19
stdev 2.83 2.49 2.44 2.00
count 194.00 194.00 194.00 194.00
min 47.00 56.00 50.00 54.00
max 59.00 68.00 62.00 63.00
difference 12.00 12.00 12.00 9.00
%different 22.64 19.35 21.43 15.38



Anyways is it possible to RMA a chip for the temperature sensors giving readings like this, should i be concerned?

Any thoughts or suggestions to further test this, or is it even possible to determine if the sensor is the issue
 

·
Junglist
Joined
·
7,233 Posts
I get this with my WC Loop too, shows one of my cores at 19 (2-3c below ambient) and the highest at 27c. I just go by the actual CPU temp.
 

·
Registered
Joined
·
430 Posts
This is normal. Most Ivy Bridge i5's and i7's have this.
Mine had up to 10C differences between cores. After de-lidding, it is down to 5C.Still, I can see the temps go below ambient in idle on one of the cores, so the sensors can not be fully trusted.
 

·
Registered
Joined
·
1,017 Posts
Everyone assumes that Intel uses sensors that are 100% accurate. Intel only intended for these sensors to be used to control thermal throttling at approximately 105C and thermal shut down at approximately 130C. Your sensors are quite capable of that so there is no need to RMA your CPU.

Intel's core temperature sensors have approximately +/- 5C of error in them and that's at the calibration point. When your 4 cores are equally loaded, a range of 10C from highest to lowest is perfectly normal. When not equally loaded and when you are a long ways from the calibration point, the range could be even higher. Your graph shows that you are right within spec.

The CPU Package temperature represents the peak core temperature. Watching that single number helps reduce the amount of sensor error that you see.

RealTemp T|I Edition
http://www.overclock.net/t/1330144/realtemp-t-i-edition
 
1 - 6 of 6 Posts
Top