• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Difference of 8 degrees between cores - why?

Soldato
Joined
10 Jul 2008
Posts
8,592
I have a q6600 and it's always been like this since I have owned it. Typical idle temps will be like this:

Core 0: 44c
Core 1: 38c
Core 2: 36c
Core 3: 42c

The CPU is lapped as is the TRUE heatsink. But, I thought that if it was down to bad lapping or mavity on the heatsink pulling one side down harder, then the pattern would be like:

Core 0: Lower
Core 1: Lower
Core 2: HIGH
Core 3: HIGH

But instead I get:

Core 0: HIGH
Core 1: Lower
Core 2: Lower
Core 3: HIGH

Have I misunderstood the core layout on the q66XX CPUs? Is it as per below?


corenumberingtp3.jpg
 
You're correct with the layout but the temp difference is nothing to worry about. My i5 has one core which is always 5*C higher than the rest regardless.
 
Indeed, 5c is understandable and fairly common, but 8c? I'm not overly bothered just thought I would check. In that case then it must just be how my chip is.
 
What is the core temp difference under load? (use OCCT to heavily stress the CPU).

Idle temps are not much use as most background programs will just use up core 0 - meaning that core 0 may be fine, it is just core 3 that is not receiving the fair share of cooling.
 
Interesting!.... I just checked temps with different apps and they seem to display the order of cores differently. i.e.

(Bold = the hot core pair)

Speedfan:
Core0: 45
Core1: 37
Core2: 38
Core3: 43

OCCT:
Core0: 45
Core1: 43
Core2: 38
Core3: 37

RealTemp:
Core0: 45
Core1: 43
Core2: 38
Core3: 37

CoreTemp:
Core0: 45
Core1: 37
Core2: 38
Core3: 43


So two display the core temps in one order and two another, which might account for this. Would you guys agree that it seems OCCT and RealTemp seem to be the correct order and associate the correct temps with the correct cores? As this would completely make more sense with regard to one side of the CPU getting more/less pressure/contact than the other side?
Perhaps it's time to lap it again? I guess I could try applying pressure on that side of the heatsink to see if temps change?
 
What is the core temp difference under load? (use OCCT to heavily stress the CPU).

Idle temps are not much use as most background programs will just use up core 0 - meaning that core 0 may be fine, it is just core 3 that is not receiving the fair share of cooling.

Under load the temps are the same but all 10c higher or so. If anything they seem to come more perfectly in line with two clear pairs, i.e.

51
51
44
44

or

51
44
44
51

depending on the monitoring software as I said above.

Cheers
 
WOW. I just tried someting quite rash. I applied a lot of pressure downwards (towards the bottom of the case) on the TRUE heatsink to see if it affected temps. It did! I also then pulled it back up towards the top of the case accross the other two cores and I could actually get it so that the temps got to within a degree or so to each other. This says to me its completely to do with how I have mounted/lapped this in some way. I found though that if I applied pressure upwards to the top of the case, that the hot pair of cores got COOLER. This is the opposite of what I would have guessed, as I thought naturally that the TRUE's weight would pull it more onto that pair. Hmm...thoughts? I managed to get the extreme of 11 degrees between the pairs when pulling the TRUE downwards, and when applying pressure upwards, I got all cores within 1-2c of each other.

Is it possible that the above architecture diagram is wrong and that the cores from top to bottom are reversed i.e.

0
1
2
3
 
Last edited:
Sounds like you may want to re-mount the heatsink (get it really tight) and re-apply thermal paste.

Personally, when I last installed a TRUE I put a washer between the motherboard mount and the back of the plate on the TRUE that interfaces with the motherboard so I could apply extra pressure, it worked well.
 
Can you talk me through what you mean with the way you mounted the true as I feel like it needs much more pressure. The screws are maxxed out, and obviously with it being lapped then the IHS of the CPU is sitting lower than it normally would as well which probably contributes to this.
 
I think thats the prob as it's lapped as well reducing the height. When I get a chance I will take it all off and pressure mod it with washers and such. Will be glad to get a more even spread and reduced temps as this may have been hindering my Overclock as its a good CPu with low vid.
 
Back
Top Bottom