Q6600 @ 3ghz Core Temp Varience

[FalcX]

Just curious if I should look at remounting my 120 xtreme or if this is normal. I have a q6600 at 3.0ghz (333x9) and in normal browsing coretemp reads the 4 cores at:
Core 0 -32
Core1 -31
Core2 -27
Core3 -34

It just seems that the diff of 7 degrees seems alot.

 

[MadScientist]

My Q6600 runs about the same as yours, a 7 degree difference at idle between the cores at 3.3 Ghz, 31, 31, 25, 32. At load, Prime95 Small FFts, the difference is only 3 degrees. My heatsink base and the IHS of my CPU are flat so I didn't lap them. I tried re-installing the heatsink several times with different TIMS and application methods and there was still a 7 degree difference. I think it just the way my CPU's IHS sits on the cores. If your heatsink and CPU IHS are flat try re-installing the heatsink. If you still get a 7 degree difference I wouldn't worry about it.

 

[BonzaiDuck]

Y'all didn' say which heatsink or cooler you're usin'.

MadScientist's notes seem to controvert what I'd concluded in empirical observation in recent weeks or months: that the statistical range of core temperatures is narrower at idle and wider at load. Hmmm. Interestin'.

Whoa!! I was wrong: FalcX is usin' the Ultra 120 Extreme!!

FalcX? Did you lap the heatsink-base? I bought mine "custom-lapped." U-120-Extreme was deliberately designed by ThermalRight with a convex ridge down the center of the base.

If you decide to lap it, I only suggest that you build a primitive jig to start grinding the ridge so that it flattens evenly on either side. I used two 1/2" x 1/2" x 6" pieces of white pine and a small clamp. I put the wood on either side of the base flush with the its bottom, and put the clamp through the heatpipes to clamp the wood to the base.

Then I started grinding, until there was a flat exposure of copper evenly distributed down the center of the base. About half-way through the lapping, I removed the clamp and sticks.

I would also lap the IHS / processor-cap / heatspreader on the processor. Take care. Tape some 400-grit wet-or-dry to a flat, glass surface. Get a latex, surgical-quality rubber glove to wear. Ground yourself, of course. Put a couple drops of water -- NO MORE -- on the sand-paper, apply the processor upside-down, and lap it slowly, carefully. Don't get any water on anything other than the IHS surface.

Also -- I suspect -- lapping the processor flat is not enough if it leaves the nickel plating on the IHS. Graysky (a member-guru here) wrote a post at another forum with pictures n' circles n' arrows on the back of each one -- showing how he lapped the IHS down to bare copper.

Nickel has a higher thermal resistance than copper, so there will definitely be an improvement in grinding off the nickel layer.

Any range of temperatures cited by either one of you for idle operation shouldn't really matter. If it spreads out under load, I'd attend to lapping and other measures. 7C-difference at load is about the highest I've experienced; 3C-difference should be cause for celebration; and at idle . . . . as I said . . . . not to worry.

 

[Thor86]

Some quads variances are higher than others, but that is normal.

 

[Amaroque]

I sometimes see up to a 7c varience on my quad. Seems odd, but it doesn't have any stability problems.