Core layout/uneven temps

thegoonden · 3 Jun 2008 at 07:43

OK, the temps are hardly tragic,
eg, under full prime95 load at 3.5GHz

65 63 63 65

Not too high (that's with a bit of a V-boost to about 1.38, but it's stable nearly .1v lower for anything but Prime, and a good bit cooler obviously).
And the difference is well within acceptable levels.

But the same chip (Q6600) and cooler (TT120, now fitted with Yate Loon "H") used to read out almost perfectly even, with perhaps a teeny delay in cores 1 and 2 getting as hot as 0 and 3

Was thinking of tweaking the cooler, but wondered what if anyone could help me guess which screw may be less tight, as I can only get at 2 or 3 of them while machine is in it's case.

If the Arrow on the chip is pointed towards the RAM, with the 2 "missing pins" top and bottom, what way are the cores laid out in space?

are 1 and 2 diagonal, horizontally side by side or vertically side by side?

Could it also be poorly applied AS5?
Should I wait til the AS5 really beds in before changing anything?
Should I really just dismiss the variance (though maybe check such cooler screws as I can reach)?

Cob · 3 Jun 2008 at 17:15

They're still very level for a Q6600. Some have as much as 12c of a difference between the cores.

Brizzles · 3 Jun 2008 at 17:27

I asked the same on another forum :

In a standard tower with you looking straight at the case the cores would be:

------------
|---- 3 ----|
|---- 2 ----|
|---- 1 ----|
|---- 0 ----|
------------

So your case exhaust would be to the left and your RAM the right of the above dodgy ASCII diagram

Your temp differences are fine btw ( I load mine up and get 10 degrees difference ).

Mark

thegoonden · 4 Jun 2008 at 00:44

Looks like it's just the concavity of the TT baseplate that's doing it then.
I know they're fine, but they used to be perfect.

It's probably because I used only the minimum AS5, whereas originally it had half a tube of Tuniq's compound on it, which must have helped reduce the problems of the concave HSF base.

Cheers.

DrSteve · 4 Jun 2008 at 09:34

Have the same thing with my Q6600. 2 of the coes are 8 degrees hotter than the other 2 at idle. Falls to a 5 degree difference under load.
I even tried lapping my chip and remounting the water block yesterday but that just lowered all the temps by the same amount. The difference is still there.

thegoonden · 4 Jun 2008 at 21:34

Oh don't misunderstand, I am not at all worried I know 2-4 degrees is so far within tolerance as to me negligible, it's just that they used to be pretty well flat.

Like I said, I think my overdose of Tuniq TIM probably compensated for the slight concavity of the TT base, mind you at the same time it made the temps a good big higher overall than they are now.