How best proceed with overheating i7-4790K?

[REnso1]

I have an i7-4790K in a GA-Z97MX Gaming 5 mobo latest F4 BIOS.

I used a Noctua NH-L12 cooler rated at 95W for cooling the processor rated at 88W in a Lian Li PC V354 with 4 fans, 2 in 2 out, case closed and case open, ambient is 27-30°C.

Temperatures in BIOS and memtest86+ were high so I decided to try stress testing and in Prime95 small FFT cores 1&2 overheated to 100°C using Core Temp.

I tried reseating the heatsink and renewing the NT-H1 TIM and opening the case but it made no difference. I have a photo of the contact pattern here.

When I tested using the OCCT benchmark I was unable to complete a test due to the processor overheating so I underclocked the processor to 3.6 GHz, disabled turbo and manually set vCore to 1.1v.

With an underclocked processor I was able to get a heating and cooling curve using the OCCT auto capture, to enable me to study the problem.

Even when underclocked the processor was reaching high temperatures, rapid fluctuations in temperature with work load suggest a bottleneck in the thermal pathway. When I tested with the intel retail cooler which came with the CPU the cooling was much less effective than the NH-L12 (even when underclocked taking just over a minute of OCCT to reach the 85°C cut off point see below) indicating the NH-L12 was doing a good job of removing heat, which meant the processor was making the heat or the source of the bottleneck.

I have discussed it http://forums.hexus.net/cpus/327593-4790k-overheating-nh-l12.html elsewhere. Advice was to contact Intel due to an absence of information relating to my retailer's testing procedures. I have asked about these but am still waiting for a reply.

So my question is how should I proceed from here? Does this qualify for an RMA? If so is it possible to negotiate this with Intel direct or do I have to go through my retailer?

I have done my best to make sure I am not doing anything wrong and I would be grateful for any pointers to any mistakes I may be making.

[JBras1]

Any idea when this will happen? We have been waiting on this for quite a while and will we ever get a reply?

What have they got so far, is someone working on it? Right now I am considering a refund if nothing happens any time soon.

The solution I have posted will make my CPU run significantly slower and even gives me framedrops in games, it will at least stop the overheating but this situation is far from ideal.

[MCout5]

I just purchased and installed this CPU and after reading many posts across the web, immediately changed BIOS settings to disable Turbo mode, as heat was indeed going up. I am disappointed that I can't use the processor to its full capacity, but at least it is stable and fast enough for what I do (mostly gaming and video rendering). My specs are as follows:

Windows 8.1 64 bit

Motherboard: GIGABYTE GA-Z97X-Gaming 5 (rev. 1.0), latest F5 BIOS installed

Core i7 4790K running @stock 4Ghz (with Coolermaster Hyper 212 HSF)

8GB (2x4GB) Patriot 1600 DDR3

Sapphire R9 290X 4GB

Crucial M500 256Gb SSD

[SElwa]

Allan, anything new on the issue, I have the exact same problem as the rest and I don't know if it's the CPU or the mobo or the stock fan I am confused!

CPU: I7-4790K (Stock Fan) L4

Mobo: GA-Z97X-UD3H (Latest Bios F7)

Ram: G.skill RipjawsX 16GB (2 x 8GB) DDR3 1866 CL9 1.5v

Full load 100% @4.4GHz or @4.00GHz ~ 95c to 100c

The stock cooler mounted correctly, I did all the stress tests and my system is stable @Vcore 1.2 but the heat makes the CPU throttling to 3999 MHz or lower

[idata]

After a lot of experimenting This is what I get:

I7-4790K Asrock Extreme6 - OEM Cooler

Set core speeds to 45X,45X,44X,43X ( can push to 46X but once a week instability.)

Set CPU Power Limit to 77 Amps. (seems to be the best temperature limiting factor)

All other asrock CPU defaults

Works nice at low stress. High stress peaks temp at 80C. But core CPU speed drops to 38X. ( Paid a lot of money for a 38X processor ! )

Is 80C (peak) a safe temp under stress?

I use the Intel Extreme Tuning Utility. I hope that it gives accurate readings, and it lets you make adjustments on the fly.

[SElwa]

@TAS3086

You mean 77 Amps or Watt?

Also 80c peak is better than mine 100c!

[SElwa]

Also, there is something very strange:

If the turbo boost is ON; under stress test @100 load the temp reaches 100c!

If the turbo boost is OFF; under stress test @100 load the temp reaches 100c!

If VCore @1.199 or 1.210 or 1.220 or 1.230 or auto (1.270 - 1.287) or even normal with offset -0.060; under stress test @100 load the temp reaches 100c!

What I mean whatever I do, nothing changes and I updated the bios to the latest version and the stock cooler is mounted properly, I don't know what is wrong, I bought that CPU days ago!

[REnso1]

ShadiElwan it sounds like the CPU is overheating in all those scenarios and the onboard sensors are throttling the clocks to prevent it passing 100°C, so it will be restricting turbo or even underclocking. This is a safety feature. It will always reach 100°C and then go no higher because of this feature. So the results you are seeing are because the retail cooling is not nearly enough for stress testing (which test?).

In normal desktop activity the CPU workload will vary rapidly and so will activate turbo only for very short periods of time, the retail heatsink is just sufficient for light use but you dont buy an I7 4790K for light use only, you buy it for its peak capability in sustained use like gaming or media processing etc. By the same token if you buy an I7 4970K you are going to want to add your own cooling because it is always going to be better than Intel retail, even on a good day. So it does not make sense for Intel to try to sell the I7 4790K based on its heatsink and its understandable that it is the minimum necessary to run Windows on the CPU.

In short you need a much better heatsink than retail to stress test this CPU, without a doubt. The CPU produces massive heat spikes and only a maximum heat gradient across the heatspreader will get it out fast enough. All the evidence suggests this is because the gap filled with thermal interface material between CPU silicon and metal heatspreader is too thick. The heatspreader is great, that is not the problem. If the gap between it and the silicon was bridged by heatspreader material instead of thermal paste the whole package would probably perform much better for very little outlay.

Dont get me wrong, IMHO the I7 4790K chip is a real technical achievement and I am grateful to have the opportunity to use it but you cant spend any time trying to optimise this chip without running up against the fact that its Achilles heel is the packaging. It would be smart to fix this for next gen as a hot CPU uses more power than a cool CPU to do exactly the same processing, by which I mean it could save money and help save the planet at the same time

[idata]

According to all the documentation that I am reading, I am setting the "CURRENT level in AMPs" limit to the cpu at 77 amps.

( I thought watts as well, but everything says amps. I can limit the turboboost limit in watts, but CPU is amps.)

The Intel extreme tuning utility (in addition to the Asrock bios), lets me control the turboboost in watts and CPU limit amps , and gives nice real time graphs, and allows real time tuning changes as well.

But still want to know if 80C is a reasonable peak Temp.

[SElwa]

boolybooly

The stress test has been done by:

1- Prime95 (64bit) - 28.5

2- Intel® Extreme Tuning Utility (Intel® XTU) version 5.1.0.23

3- IntelBurnTest 2.54

4- AIDA64 Extreme Trial version

So do you mean that my CPU is not faulty and my board GA-Z97X-UD3H (rev. 1.0) is not faulty too? And the real problem here is the stock cooler?

If the stock cooler is the real problem here, then I'd be relieved, but I have a little concern if the stock cooler makes my CPU reaching 100c under the stress test, then from what I heard and understand that most average aftermarket cooler will make the CPU reaches 70c - 85c under stress test (-30c to -15c difference in temperature) and that is not good you know why? Cause then I can't overclock the CPU beyond it's turbo frequency and if I am lucky I can get it to 4.5GHz do you think that would be normal? It's K CPU if I can't overclock it then why I bought it in the first place?

Lets assume the stock cooler is the problem which aftermarket coolers from that list you suggest to me (that is the available to me right now to choose from):

- Thermaltake Water 3.0 Performer CPU Cooler

- Cooler Master V6GT CPU cooler

- Thermaltake NiC F4 120mm CPU Cooler

- Thermaltake Water 3.0 Pro CPU Cooler

- Thermaltake Water 3.0 Extreme CPU Cooler

- Cooler Master Hyper 212 Evo

- ZALMAN Reserator 3 MAX Ultimate Water/Liquid CPU Cooler

[SElwa]

80c peak is safe? Yes, that CPU can handle it. Is it reasonable or normal? NO and NO!

[REnso1]

ShadiElwan logic suggests you cannot be sure of the CPU or the motherboard until you have a cooler which can perform a stress test you can compare with other peoples results. The Intel retail cannot even run the Intel Processor Diagnostic Utility test without reaching 100°C. You can be sure that once you have verified the components are good you will still need a good cooler to get the best out of the CPU. So it is worth getting a better cooler IMHO.

I only have direct experience of the NH-L12 with this CPU which is just a bit less than adequate, so its hard to advise you on cooler choice with any confidence, maybe others can add to this. Best advice I can come up with is this, selecting a cooler is a personal choice as it depends on how much space you have in the case, how heavy you want it, noise tolerance etc. My advice would be to look at recent cooler review comparisons (just google the cooler name and add "review" to the end) and select the one which will fit your needs which provides the best performance. Make sure you look inside your case though, (safety tip for anyone inexperienced opening a case - always switch off at mains and wear a static strap) for air coolers get a tape measure and physically check the space you have for the cooler height and width and RAM clearance to compare to manufacturers specifications, for water coolers check the radiator dimension against case fan openings and any places you can put it without bumping into other stuff, ideally use fans in a push+pull configuration on the 120mm radiators but one set of fans is usually adequate on larger rads. Then you should take an hour or two to google for people reporting issues with that cooler and cooler/RAM/motherboard/case combinations etc first.

All the coolers you have listed look good in the reviews I saw.

Once you have good cooling, if you read the suggestions in this thread you will find there are a few essential tweaks which can reduce temperatures while maintaining as much performance as possible. (eg I summarised some tweaks in the "/message/253580# 253580 correct answer" post FYI, but other contributors have also made suggestions.)

Hope that helps.

[SElwa]

I ran Intel Processor Diagnostic Utility test and ironically it passed the test! Whatever, but here is the thing: It gave me that warning

-------------------------------------------------------------------------------------------------------------------------

Detected CPU Frequency is --> 4.40064

CPU Frequency Test Passed with Warning!!!

The processor frequency could be affected by power management features. In order

to ensure the processor frequency is not affected by a power management setting,

enter BIOS, load defaults, and disable any 'Over-Clocking' or power

management features that may affect processor frequency. Also, set any

software applications which control power management features to default.

-------------------------------------------------------------------------------------------------------------------------

What does that mean exactly? And another thing, you said you set "CPU Watt Limit" in bios to 110! Is that safe? If yes, what is the maximum safe allowed Watt/Amp for that CPU considering the VCore is 1.2100v?

[REnso1]

The safety depends on the temps which depend on the cooler and the load.

If you try "http://www.alcpu.com/CoreTemp/ CoreTemp" you can watch the relationship between CPUclock / watts / temps and decide for yourself.

[PZWAG]

My 5960 power consumption goes up 20 watts after almost exactly 3.5 minutes on prime95 version 28.5 Thats with cmos defaults 3 ghz

Shown is prime95 and intel stress test set to low priory. The problem is not apparent with only intel stress test. Is this normal?

Im aware power is very high on prime95 versions 28.5 giving very high current, but the 3.5 minutes issue ive not heard about.

If overclocked cpu 1.09 volts 4000 mhz, it can jump up 80 watts. At this point things get hot and or crash

[PZWAG]

Can anyone tell me where karwos thermal test program url is?

the url doesnt work

http://www.cyclone-networks.eu/IntelThermalTest.rar http://www.cyclone-networks.eu/IntelThermalTest.rar

[APark7]

Just wanted to chime in with another data point, as I'm experiencing this issue with a new build as well. Here's the specs of what I've just built:

OS: Windows 7 64-bit

MB: MSI Z97-GAMING-5

CPU: Core i7-4790K with no overclocking running a MASSCOOL 8W2002B1M4 CPU fan (probably only marginally, if at all, better than a stock cooler)

RAM: 8GB (2 x 4GB) G-Skill Ripjaws X Series DDR3 1600

GPU: MSI GTX 970 GAMING 4G

First off, I don't intend to overclock this CPU (at least right now) . I opted for a CPU cooler that wasn't stock only for ease of installation. But I did want the flexibility to possibly explore this down the road at a later time. Unfortunately, any kind of stress testing using a variety of programs (Prime95 v26 and v28, OCCT, etc) results in core temps upwards of 95-100C in most cores without any overclock (aside from the the turbo boosting that happens stock). I've reapplied thermal paste multiple times just in case to verify the install with absolutely no change whatsoever. I was also able to get a replacement CPU from Amazon.com, but it exhibited the same behavior (if not slightly warmer actually). I'll probably end up sending the new one back since it seems to be marginally even worse. I can post some OCCT results later tonight if it would be useful.

In an effort to remedy the issue, I ordered a Scythe Mugen 4 as suggested further up the thread, but it unfortunately won't fit in my build because the fan interferes with the RAM heatsinks on my Ripjaws. It's going back, and I've got a Cooler Master Hyper 212 EVO on order now that looks like it should clear the RAM better. We'll see what kind of difference that makes when I receive it tomorrow.

But this whole scenario is really pathetic. The whole point of buying a K "unlocked" CPU is to have the flexibility to overclock with some better-than-standard cooling. But it seems these CPUs (or at least a certain batch of them), needs much better-than-standard cooling just to operate under their normal thermal requirements! I'm tempted to return the CPU, but I'm not sure what I'd get in its place. Do the locked i7-4790 CPUs suffer from the same troubles? I've never really been into overclocking (and honestly know very little about it), so if swapping for a locked CPU makes this problem go away, I might just consider it.

[SBuck1]

Another data point. I just built a 4790k machine on a gigabyte GA-Z97X-SOC mobo. Using a cooler master 212 EVO cooler, it idles around 35C, and trying small FFT in prime95 results in an immediate reboot. I made some of the adjustments another user suggested a page back, ie manually setting vcore to 1.18v, turning off turbo etc. Now I can at least run prime95 without immediately rebooting. Temps still shoot way into the 80s though.

[APark7]

That is really disappointing to hear. I was really hoping the Hyper 212 EVO I have on order might at least get me by until Intel has something else to say about this.

While I appreciate the fact that allan_intel is looking into this problem for us, has anyone called Intel and opened an actual support ticket for this issue? I'd be curious to see how that would play out. I may do that this evening.

[SBuck1]

I will say that my thermal paste is not the greatest, it's that antec stuff. I have some proper AS5 on the way, which I imagine will help a little, but it's not going to stop these unbelievable temp spikes.

[APark7]

Yeah I doubt that's going to make much of a difference. I've been using AS5 all along (admittedly with a cooler that is only likely marginally better than stock). But you're right, it will make a small dent in these absurd numbers at best.

[SBuck1]

At this point I think I'm probably just going to RMA the processor. I've eliminated enough of the variables that I'm confident there's a CPU problem. It's not the only problem certainly, also going to talk to Gigabyte about why the heck their board is supplying almost 1.4v to my processor under the slightest load, but even with the VCORE set to a fixed, sane number these issues are still alarmingly bad. This was actually my first ever Intel build, I've always used AMD in the past. Can't say I've got a great taste in my mouth thus far.