i7 2600K with Cooler Master 212 Evo [High temps]

[hunter45]

So I wanted to OC my 2600K which is why I bought a CM 212 evo.
I installed it, and on stock settings with prime95 I get a maximum core temperature of 64C. I did read online a lot about the temps people get and this is way to high. With this CPU and cooler, most people get 65C max core temperature on prime95 load WITH a 4.5Ghz OC so something is clearly wrong here.

I have tried already to reseat the cooler 4 times. The first time I used the pea method to thermal paste. The second time I used the dot method. The third time I did what the cooler master instructions said to do (spread a thin layer on the cooler bottom) using a credit card. The final time I tried the dot method once more seeing as this is the one many people advise to do. Every time, the temps were pretty much the same - 64C on the highest core.

This is really annoying me. I tried resetting the BIOS, I even manually measured the Vcore on board with a volt meter and it is correct. Also added a second fan (Noctua P12) as pull and it made a minimal difference (about 2C)

Can anyone guess what could be wrong?

The case I have is the Corsair Obsidian 550D. All fan bays except one are populated.
Motherboard is Gigabyte P67A-UD3R-B3

 

[Charlie98]

Reduce your vCore at stock, that's what I had to do to cool off my 2500K. I'm running a 4.1GHz OC with a .07 undervolt.

 

[hunter45]

Reduce your vCore at stock, that's what I had to do to cool off my 2500K. I'm running a 4.1GHz OC with a .07 undervolt.

I tried a typical OC to 4.5Ghz, with 1.3V vCore. This resulted in a max core temp of 75C. Still high, its supposed to be about 65Cish.

 

[BonzaiDuck]

I tried a typical OC to 4.5Ghz, with 1.3V vCore. This resulted in a max core temp of 75C. Still high, its supposed to be about 65Cish.

Your expectations about Prime95 (assuming sFFT test) could be based on information coming from an earlier version of the software, while you're using a later version.

But Prime95 should still give temperatures lower than Intel Burn Test or affinitized LinX with the AVX extensions.

I could suggest that the Obsidian 550 case could use some attention to intake ventilation. Lapping the processor IHS to bare copper and using diamond-paste TIM could be worth about 4C improvement. Ducting the 212 EVO to the exhaust with a beefier exhaust fan could be worth 5C - maybe a little more with the EVO.

Even with these considerations and assuming you attended to them all, I'd expect your 212 EVO to perform 5C worse than my D14 (2600K @ 4.6) and 10+C worse than my EVGA ACX cooler (2700K @ 4.7).

So if Prime95 sFFT gives me 68C (room-ambient 77F) with the D14, you should be able to get 73C. But that would be with drooped, load voltage of ~1.35+. So I still have to wonder what's going on with your system of 4.5Ghz and 1.30V.

 

[hunter45]

Your expectations about Prime95 (assuming sFFT test) could be based on information coming from an earlier version of the software, while you're using a later version.

But Prime95 should still give temperatures lower than Intel Burn Test or affinitized LinX with the AVX extensions.

I could suggest that the Obsidian 550 case could use some attention to intake ventilation. Lapping the processor IHS to bare copper and using diamond-paste TIM could be worth about 4C improvement. Ducting the 212 EVO to the exhaust with a beefier exhaust fan could be worth 5C - maybe a little more with the EVO.

Even with these considerations and assuming you attended to them all, I'd expect your 212 EVO to perform 5C worse than my D14 (2600K @ 4.6) and 10+C worse than my EVGA ACX cooler (2700K @ 4.7).

So if Prime95 sFFT gives me 68C (room-ambient 77F) with the D14, you should be able to get 73C. But that would be with drooped, load voltage of ~1.35+. So I still have to wonder what's going on with your system of 4.5Ghz and 1.30V.

Thanks for all of your suggestions. Currently, my case has 3 intake fans (2 front and 1 side) and 3 exhaust fans(2 top and 1 back). The 212 Evo has a push-pull config.
However I have noticed something quite weird - I left prime95 on for about an hour and came back and touched the heatsink. It was barely warm. The top of the heatsink was cold. Next I touched the copper pipes and these were a little warm. Finally, I touched the metal base and this was just warm. I then felt the air the pull fan was exhausting and this was just a little warm. It leads me to believe maybe the copper pipes have not been
installed properly during manufacturing. Could this be the reason?

Found someone that received 212 with faulty copper pipe: http://www.tomshardware.com/forum/271386-29-defective-hyper
His temps were also bad like mine, replacement fixed it.

Is a 212 supposed to have a hot heatsink and/or copper pipes on full load??

 

[flexy]

* cooler not sitting right due to warped board, missing backplate, over-tightened screws or something else wrong with your mounting?

* bios setting with adaptive vcore on or similar (rather than manual vcore), adding voltage to your cpu

* defective HSF? (IMHO least likely)

* CPU TIM spread underneath (inside) IHS bad or especially poorly spread?

Before checking into hardware faults, in my order of likelihood:

1) bios setting
2) HSF not flush

 

[hunter45]

* cooler not sitting right due to warped board, missing backplate, over-tightened screws or something else wrong with your mounting?

* bios setting with adaptive vcore on or similar (rather than manual vcore), adding voltage to your cpu

* defective HSF? (IMHO least likely)

* CPU TIM spread underneath (inside) IHS bad or especially poorly spread?

Before checking into hardware faults, in my order of likelihood:

1) bios setting
2) HSF not flush

1) Vcore is manually set in the bios to 1.3V for 4.5Ghz. Like I said, I measuered that with a multimeter straight from the board and it is correct.

2) I have reseated the cooler 4 times now. When I try to move it, it is secure. I also tightened screws all the way down. For thermal pasting, I used the guides online and at the moment have the pea method applied. I used exactly the same amount as in the guides.

The thing is, how are the temps off so badly? They are like 15C off. My TIM application has to be pretty bad for that to happen. Also everytime I have reapplied I get pretty much the same temps.

 

[Charlie98]

What kind of TIM are you using?

I still think you are using too much voltage... everyone seems to think the first thing you have to do on an OC is pump the voltage up. Voltage = heat. When I first started OC'ing my 2500K, I had temps in the 90's! ... with a 212+. I even added a second fan and built a duct for it. The ONE thing that brought my temps down was lowering the vCore. I still managed a decent OC... 4.1GHz stable at -.07v which was all my Gigabyte board was capable of. OP, your board is in the same family as mine, and I can tell you some of my OC limits are because of the board. I dropped the same chip into an ASUS Z68 board... and I was immediately able to get another .2GHz using the same settings... with lower temps. No fooling... and I wasn't even trying to push it. My Gigabyte board would have BSOD'd by then.

Unless you are using some weird TIM, the 212 EVO is a better cooler than the 212+.

As far as heat sink temperature... hot to the touch... think about it: The heat generated by the CPU is dispersed over a 1"x1" square of metal (CPU lid) and then transferred to a gigantic (by comparison) metal appliance that disperses that heat throughout... of course it's not going to be hot to the touch.

 

[Dasa2]

you have a budget hsf with that oc i wouldnt be surprised if you hit over 80c while stress testing
provided it stays under 80c during normal use i would be happy
on custom water under load im looking at 50-60c for my 2600k

1.35v for 4.5ghz sounds about right for a 2600k i run ~1.3v for 4.3ghz and ~1.4v for 4.9ghz

 

[MontyAC]

I had a defective 212+ and returned it, the heat pipes didn't cool properly. The replacement one works fine.

 

[tweakboy]

Nice ending to your story monty very nice.. gb