Struggling with heat generated from i7-4790k

[Slingfox]

I have been struggling with the heat generated from my i7-4790k for around a week now, I decided to go and buy a couple of different TIM's and a couple of new heatsinks yet when running Prime95 on Large FFT for around ten minutes the temperatures hit Tj Max and the processor starts to downclock.

I haven't overclocked it, I'm running with default settings and have updated the BIOS, the motherboard is an Asus Maximus VII GENE Z97 mATX, the current heatsink is a Phanteks PH-TC12DX with two Enermax Twister Pressure 120mm fans in Push/Pull configuration and each of them can handle 80+ CFM and a relatively high static pressure.

I decided to try a Corsair H75 AIO hydro cooler, and when it came to idle temps it did excel, but when testing with Prime95 it fared much worse than the Phanteks.

The idle temp with the Phanteks is around 36 - 39c, load temps with Prime95 are around 91c for the majority but after ten minutes we're at around 101c at which point I end up stopping the test, with the Corsair H75 the idle temps were around 31c but load temps hit 100c much quicker than the Phanteks.

As for airflow in the case, I have a 230mm fan bringing cold air in, the push/pull heatsink fans, two Enermax Twister Pressure fans at the top exhausting the hot rising air, and a fan at the back of the case bringing in more cool air over the heatsink.

I have tried Arctic Silver 5, IC Diamond 7 and as of today, Prolimatech PK-3, using the grain of rice method and I have checked a couple of times by removing the heatsink just to ensure the TIM is covering the CPU fully, which it is.

I'm completely out of ideas, and I know Prime95 is a tool specifically for pushing temps to the max, but I've never seen temps like it from anyone else, ambient temp in here is around 23c.

Any ideas?

[GoNz0]

RMA it, thats way too high with that cooling.

[Slingfox]

I'll send Overclockers an email, only had it a week, I'll see what they say.

[Biscuit]

You're using aftermarket and overheating?! Man if you're overheating with STOCK cooling, there is something very wrong.

Let us know how things progress.

[HalloweenJack]

4770k thermal shut down on stock cooling in a sealed case so this isn't a surprise - these things run hotter than ever then AMD 9590!

[kompukare]

Don't most Haswell BIOS's have this feature set where the CPU gets a 0.1V boost when running AVX/AVX2 code? Or even that some motherboards run the turbo bins at the max setting (in effect overclocked) even at default?

While it sounds like the OP's CPU is faulty (although AVX2 hype and reality seem very far apart especially with Intel's supplied stock cooler), monitoring the voltages during stress testing might say something. Or even the turbo modes.

[Slingfox]

Thanks for the input thus far guys, I know a couple of people have mentioned stock cooling, I am not using the stock cooler, I am using decent aftermarket coolers.

I have ordered a new i7-4790k from another retailer, Overclockers have accepted the RMA pending further testing by their engineers, so I'll be shipping that back to them on Monday.

I tried manually fixing the voltage and turning off all the fancy ASUS overclocking tools today, set the multiplier to x40 thus setting it as the standard 4GHz clock and the temps didn't hit 80c, as soon as it ran at Turbo speed, aka x44 was when it hit 91c fairly quick.

It really annoys me due to the fact that my buddy is running the same setup and his temps never exceed 70c with similar if not worse cooling than I have so there is certainly an issue here somewhere with my build.

[Bluecube]

Are your coolers secure and fitted correctly? I'm sure they are but I had a case where the CPU temps were a bit out of the ordinary, going waaaaay high when stressed. Turned out that one of the HSF pins hadn't clicked in place properly, something that wasn't obvious. The same thing happened when I got a H60i water cooler - the problem was I had put it in the wrong way round!

[Slingfox]

I did hear something about the Corsair AIO Hydro series needing to be installed in a particular way so I tried two of the possible ways and the temps were pretty similar, as for the Phanteks, it's as secure as it can possibly be, it doesn't rotate and there is no give/wobble when trying to move it with a gentle prod, I've taken both coolers off numerous times due to paranoia of the TIM not spreading correctly but each time it has been absolutely fine.

It's one of those cases where I wish it was just me messing up, but I've built so many machines to date that I know right from wrong, and thus why I'm so stumped with this particular issue.

[Slingfox]

Just ran OCCT, Linpack 10 minute test with AVX, purely to gather evidence for the RMA, along with photos of the machine.



This is what happens with ten minutes of Prime 95 (Large FFT).

[Slingfox]

I received a new chip today, ordered it from eBuyer over the weekend and just finished testing it in the machine and the difference is night and day, maximum temps from the new chip, with Prime95, OCCT and AIDA64 are around 72c.

I just hooked up the Corsair H75 and temps maxed out at 71c although it sat at around 68c most of the time, I checked the batch numbers between the suspected faulty chip and the new chip and they are from different batches which is reassuring, seems I lost out on the silicon lottery with the first chip, it just ran super hot.

[GoNz0]

good to hear, better present that to OCUK if they reject your RMA!

[rupertsilva10]

Check first your coolers if they are properly fit and the fan is spinning properly. Check the cooling fan location if there are any obstacles that the air cant pass through properly.

[HalloweenJack]

71 is still hot oO - my 9590 ran at 57 under prime....... seems intel chips make good space heaters now!

[MrJim]

Were you allowing the motherboard to set automatic Vcore voltages? I know my Asus motherboard sets Vcore way too high on auto, so it might be the same for you?

[Slingfox]

Were you allowing the motherboard to set automatic Vcore voltages? I know my Asus motherboard sets Vcore way too high on auto, so it might be the same for you?

On the first boot yes, but then I started to set the voltage manually until I found a stable voltage, my current working 4790k is behaving itself nicely on 1.1v @ 4.4GHz whereas the other chip I have since RMA'd was running stupidly hot even at 1.1v, I even turned turbo off to bring it down to 4.0GHz which ran nicely but temps still spiked into the low 90s, only when I limited the multiplier to x35 did the temps start to fall into mid 70's.

I guess it's just a case of waiting for OCUK to give me the verdict, I sent the chip back to them yesterday and they've informed me that they have the chip and are currently testing it.