Massive difference between core temperatures

[El_Roberto]

I just bought 3 new 110cfm fans for my pc and after installing them I noticed that there is a massive difference between core temperatures in my Q6600, Also the temperatures are extremely high for such a mild overclock (3Ghz with 2.268v). Idle the cores are at 46, 44, 35, 35 but at load there is a 17c difference between cores - 73, 72, 55, 55. Could the cpu be failing (it was bought 2-3 years ago) and when I got it I could run it at 3.0 ghz stock voltage (1.212v) now I want to put It back up to 3.2Ghz but I cant raise the voltage anymore becasue the temperature causes core 0 and 1 to fail at prime 95. I noticed that the heatsink (TRUE) is cold to the touch so I reseated it and it dropped the temp to what is listed above but it is still cold - could it need more thermal paste?

Rig:
- Intel Q6600 with TRUE (110Cfm Scythe slip stream)
-GSkill 4GB DDR2 1066Mhz
-Gigabyte GA-G33M-DS2R
-Asus GTX285
-500Gb 7200.11
-Corsair 520HX 520Watt
-CoolerMaster CM690

 

[Slaymate]

Have you lapped your True? I haven't seen one yet that didn't need it. And you should also consider lapping the cpu.

 

[El_Roberto]

It hasn't been Lapped but it is the same cooler the system has allways run so it doesnt make sense that the temps are so high. Thanks for the videos I will try to lapp it this week to get better temps but I doubt that is what is causing problems.

 

[Burn]

(3Ghz with 2.268v)

You mean 1.268 right? Pretty sure that the chip would be dead with 2.268 volts.

What thermal paste are you using?

And I agree, while lapping does help performance (especially on TRUE's) THAT much of a difference is ridiculous IMO.

 

[El_Roberto]

I am using Manhattan branded silver thermal grease. And yes 1.268 is correct

 

[Slaymate]

The base of the True has a curvature to it. Because of the curvature the True may have warped the intergrated heatspreader of the cpu. If you can match the curvature with the warpage just right your temps may return to normal. But that is all speculation.

 

[philbar71]

My q9300's cores sit a little weird too. Core #2 idles about a good 10C colder than the other cores, I think in my case it's a stuck temp sensor. Not a big deal.

 

[El_Roberto]

That makes sense Slaymate, Yea the 45nm Core 2's are terrible for stuck sensors. I ran the sensor test in real temp to check mine and they all show up fine I think my problem is what Slaymate said. But has anyone else noticed their TRUE barely heating up? I would think even with problems it would be noticably warm.

Edit: I am definately going to Lapp it now, I googled lapping TRUE and with the "washer mod" people where getting around 50c load temps on 3.2Ghz Q6600 - Makes mine seem even more terrible.

 

[Slaymate]

If the base of the heatsink is only touching a small amount of the heatspreader then the Thermal Transfer is going to be low. If the Thermal Transfer is low then the heatsink isn't going to get very hot

 

[El_Roberto]

Well I done the washer mod and the temps dropped 12c! Now im going to the hardware store tomorrow to get sand paper to Lap it. Also the core temperatures only have a 14c difference now.

Edit: Turns out I must have accidentally shorted the cmos reset pins as it was only running 2.4 at 3.0 it dropped the temps by 5-7c per core