Overclocking two 3570Ks, 10C+ difference

[Kaisenn]

Hey everyone, I am a first time overclocker and running into a curious issue I was hoping to get some feedback on.

I just built my girlfriend and I a pair of almost identical systems, both running 3570K, Gigabyte Z77X-UD5H, Corsair DDR3-1600 4x4gb, with Corsair H80 CPU coolers. When I started to overclock them, I noticed hers runs at idle about 5-6C higher than mine. I got my system to 4.2GHz at fixed 1.15V with LLC on Turbo (second highest setting on the Gigabyte MBs) (no other adjustments from optimized defaults) and this is when i first crossed 60C under load with Prime95 Small FFT.

Hers on the other hand I have set right now at 4.0GHz 1.11V LLC Turbo but Prime95 will break 70C on all 4 cores. I have a CM ATCS 840 to her Corsair 400R so I expected some difference, but not this drastic. Ambient is the same, both cases are about 5 feet apart. Mine is in stock fans + the optional bottom 120mm intake. Hers is the stock front 120mm intakes + two 120mm exhausts in the top (PSU blocks the bottom mount)

Is this 'normal' in the variance from CPU to CPU + my full tower to her mid? Or is there something I can look into to get her temps down. I have reseated her water block and from what I can tell I am getting good contact.

Thanks for the input.

 

[ShintaiDK]

Its normal. 2 chips aint equal.

 

[Ferzerp]

Also, "idle" temp readings are meaningless (at least on intel chips). You don't really think that some cores idle below ambient do you?

The only meaningful (or rather comparable) temp readings are synthetic full loads. Also, full load from app x is not comparable to full load on app y.

 

[rgallant]

batch #'s for ref. just in case of a big swing . week 6 vs week 22

 

[sm625]

I would like to know what your temps will be if you swap the two cpus. You would think the temperature difference would follow the cpu, but it would be nice to verify that. If so, I would delid the hot one and redo the TIM.

 

[bl00tdi]

I had been reading that lower VID ivy bridge chips can be a little "leaky" as is being reflected in your temps. The bottom line is, not all processors are created equal as was already mentioned.

 

[BonzaiDuck]

It could be any or all of the reasons people cite above, but also the issue of TIM between the IHS and CPU-die. One can deduce that use of a paste would have a manufacturing process with greater variability than the previous method in which Intel used the indium fluxless solder. [I invite anyone to visit the appropriate thread -- not hard to find on this forum: "De-lidded my i7-3770K . . . "]

 

[Kaisenn]

I did end up swapping the processors last night just to see what would happen, and the temperatures mostly followed. My better cooling system did help a couple of degrees but the lower voltage one is still showing up at a much higher temperature.

I read the delidding post too and it looked interesting, but for this particular project I think its unnecessary. I got my 3570K humming along nicely at 4.5GHz with temps between 68 and 75 max on its stress tests, passed the overnight stability test, and I can live with that. And my girlfriend isn't enough of a power user that more than 4.0GHz matters to her.

Thanks for the input, and any other feedback is welcome.