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.

[cpalm3]

So in looking at Gigabyte's screenshot, it looks like they hit a max of 88 degrees celsius at 100% load, and they deem this as acceptable?? My old i7 940 never hit 60 degrees at 100% load. I think this 4790k should be able to run at stock turbo speeds of 4.4ghz without overheating, but we need an official answer from Intel on what are the acceptable temps at stock speeds.

[REnso1]

Good point chilipalm, I believe the package thermal envelope is 72 - 80°C. I am going by a well researched http://www.tomshardware.co.uk/forum/id-1800828/intel-temperature-guide.html guide on Tom's Hardware forums. This states that temps above 80°C are injurious to the processor due to migration etc.

However my own setups temps are higher than Gigabyte's.

This is a screen report from testing with the Intel HS/F with NT-H1 and fan set to full speed in the BIOS which is otherwise optimised defaults apart from SATA which is set to RAID due to an issue with Win7Ultimate not seeing the Crucial M550 1Tb SSD in ACPI mode. So DDR is SPD 1333 1.5v and not XMP. Even so it overheats after 10 mins of AIDAx64. I have sent this to Gigabyte along with precise replication instructions so they can match my BIOS settings exactly.

[Allan_J_Intel1]

@chilipalm and @TAS3086, I have forwarded your question and information to our engineering department.

Thanks

Allan.

[cpalm3]

Allan, have you heard back yet from the engineers the temperature range that is acceptable for this processor at 100% load?

[Allan_J_Intel1]

I am still waiting for this type of information. I appreciate your patience.

Allan.

[AStar11]

Hi,

I am having a similar issue with my 4790K and Asus Maximus VII Formula.

I am using an H100i. My first attempt with the cooler and the supplied paste (stock BIOS settings, all AUTO, no OC) resulted in 100C temperatures in CoreTemp with Prime95 but the system didn't shut down. I thought the cooler wasn't applied properly so I also put washers behind the backplate and re-applied using Thermalright Chillfactor III thermal paste. The issue persists and it is getting far too hot to be normal.

I also purchased my CPU and Motherboard from SCAN, so should I get in contact with them?

Thanks.

[REnso1]

Sponge, that is your call. SCAN will pay for pickup and redelivery if they find a fault. If they dont they will charge you at least £30.

If you do RMA, I advise you to insist that they test the CPU with a test which can detect overheating such as AIDAx64 which I know they have available to use. They are willing to authorise on the basis of a test using AIDAx64 and Intel retail HS/F, if the AIDAx64 processor graph shows a red thermal throttling indicator (and CPU temps reach 100°C) then that is sufficient grounds. But you should insist that they use AIDAx64 themselves.

I say this because in my last RMA despite stating that their bench test was AIDAx64 they tested the CPU with Intel Processor Diagnostic Tool only and then returned the CPU to me with a new motherboard whereupon the overheating persisted. I have checked and discovered that this tool does not detect overheating effectively, since a processor core can reach 100°C during the test (which is throttling temperature in AIDx64) and yet still pass the test. Please see the following screenshot for evidence of this.

[AStar11]

I have done some tests and have some interesting insight...

The overheating only seems to occur with Synthetic benchmarks such as Prime95. Under games it is fine, and even under 100% load with image editing software such as Paint.net. Since all I use my PC for is gaming and Image editing (mostly) I have decided not to RMA the part.

It's clearly not a packaging defect: since it would get immensely hot regardless of the workload. Part of me thinks that when running special instruction sets such as FMA3 and/or AVX2 at full Turbo speeds, the chip can't get the heat off quick enough since these units are larger and require more power to use.

Regarding the motherboard, my Formula wasn't allowing my chip to thermally throttle, which is why it hits 100C+ in these tests. I tested it in an MSI Z97 PC mate under the same test with the included stock cooler, and it rises from 70C to 100C+ where it throttles down to 3.4 GHz to keep cool enough, sitting at 99-100C.

The latest iteration of Prime95 uses FMA3, AFAIK, not sure about the Intel Tool or AIDAx64, but it's interesting. This is still something to bring up to Intel, though, and I'm surprised it's not more widespread. Either way, since my chip remains in the mid-low 40s during gaming, and peaks at ~52C under 100% load image editing on my H100i, I decided not to RMA.

Hope this helps.

[idata]

Passable solution to my problem:

Set all core and cache ratios to x45. (gets real hot if they do not match)

turbo is activated.

Set Max current levels to 82 watts

Under stress, temp will hold high 60's to low 70's. I think this is acceptable.

Increasing watts to 83 will jump to 90's , so I live good and not too hot at 45000 and 82 watts.

Can not run all cores at x46, I think just setting a few may give me a few extra cycles.

system will peak at 4500 for bursts of time, but under severe stress, it will cycle back to 3800-3900 as the temp gets hotter.

starving the cores of power seems to be good for me, I wish that there were some other algorithmic way.

Comments appreciated,

[REnso1]

I had to buy AIDAx64 Extreme as the trial period ran out so I am sticking with it for a year at least! It has good diagnostic sub menus.There seems to be some debate about the latest version of Prime95 heat Sponge, which is why I am using AIDAx64 as a benchmark which SCAN recommended. Its not as extreme as Prime95 but it does detect and notify thermal throttling so its a fair bench test to compare like for like with Intel retail HS/F and if AIDAx64 throttles you know there is something wrong.

SCAN's return costs are set up to put an onus on customers to make sure they have tested properly and rechecked basic things like cooler seating etc. Not wishing to seem a totally spoiled brat customer I accept I have to take some responsibility, though I hope that doesn't put people with legitimate problems but less confidence off getting proper support.

I have been down the power throttling route TAS3086 and it works, that trick is in reserve for when the platform is good at default settings. As Gigabyte kindly demonstrated it should be possible to run AIDAx64 continuously on an I7-4790K at default settings using the Intel heatsink fan on this motherboard without thermal throttling. The set up I have can't do that so CPU and motherboard are being RMA'd today for reevaluation. I will update this thread to say how that goes.

[cpalm3]

At what point can you rule out the cpu cooler? The tech support where I bought my computer told me the cooling unit is the problem. I use a Corsair H60. The unit keeps my cpu at around 39 degrees celsius idle. If I run anything any application say consistently at 15-20% load, temps increase to around 60 degrees or so. There is one game for some strange reason uses the CPU a lot to where it reaches in the 90s.

[REnso1]

UPDATE : I have been contacted by my retailer SCAN (by phone) to say replication of this overheating problem was successful, not only this but when they went on to test replacement CPUs using AIDAx64 and Intel retail HS/F they said they discovered a bad batch of three overheating 4790ks before the fourth CPU proved to work within spec under the same conditions.

Here is their screeny showing a CPU operating within spec at 4.4 turbo, running AIDAx64 under the retail Intel HS/F.

EDIT further discussion revealed this was made using a http://www.zalman.com/global/product/Product_Read.php?Idx=164 Zalman CNPS9700, not the Intel retail Cooler FYI.

So these CPUs will be returned and I hope they will stimulate Intel to consider its manufacturing process and package design. I hope this will be useful information for others who have thermal issues. I would suggest you could test using AIDAx64 and the Intel HS/F as a standard to compare like for like and if your CPU causes thermal throttling under those conditions you should probably RMA it and get one which doesn't.

[cpalm3]

So it looks like if we're going over 85 degrees max on 100% load, we have a bad CPU. I decided to try the Corsair Hydro H100i and now hit a max of 93 degrees instead of 100. I'm going to see if I can RMA my processor. Intel really needs better QC.

[REnso1]

Further update, I received the replacements back and they ran hot in AIDAx64 but just under 100°C throttle temps with the Intel HS/F.

So I checked how the screeny was made, apparently the 82°C screeny was made using a http://www.zalman.com/global/product/Product_Read.php?Idx=164 Zalman CNPS9700 cooler, not the Intel HS/F. So I was mistaken in assuming they used the Intel cooler, that was because they told me to use it so I assumed it was a standard bench method. I will edit my previous post to reflect that, sorry if it mislead anyone, I am in the middle of testing but I thought I better make this clear the moment I knew about it.

So to summarise FYI. Currently the situation is that my retailer accepted my RMA based on Intel HS/F AIDAx64 stress test (on their instruction) and agreed it was overheating after testing for themselves.

They then replaced the components but used a non Intel cooler to test these before sending them.

They are also recommending when running the AIDAx64 test that a better than Intel cooler is required, to prevent overheating for this model of CPU (4790K).

So while an Intel cooler can be used for testing it is not the recommended cooler for stress testing and a thermal throttle event in AIDAx64 using Intel HS/F does not guarantee an RMA.

I have tested with the Intel HS/F and the retailer tested replacement CPU does suffer thermal throttling in AIDAx64 at default settings with 4.2 turbo. I am further testing this with a Noctua NH L12 and will attempt to replicate the screeny above.

@ chilipalm, throttle happens in AIDAx64 at 100°C. So wrt to 85°C I don't think that is the right value, and it also depends on what heatsink you are using and which test (Prime95 latest is excessive) and what I have read suggests its important to check seating first, in case like Sponge you might find the cooling improves if you get it seated better. Also wrt agreeing an RMA in the end it is between you and your retailer, I expect different retailers have different policies. That is all I can think of to say on that.

[ATama2]

Hi,

I have similar issue with Intel Core I7-4790K and Gigabyte z97-d3h

cpu hit 100 degrees cel at 100% load after 1 second with prime 95 v 28.5

(Core I7-4790k temperature problem )

[AStar11]

Don't use the latest version of Prime95, use Version 26.6 and try it again, see if the temperatures are normal?

The latest version uses FMA and is very stressful on Haswell processors. I had a similar issue with my 4790K and Prime95 v28.

[AStar11]

I seem to have solved my problem. But I'm not entirely sure how. I'm now fairly sure my chip is actually 100% fine, since I have done 12 hour runs of Prime95 (v26) , multiple runs of Intel Burn Test, and at least an hour of AIDA64 Extreme, and the temperatures are peaking at 59-60 *C at full 4.4 GHz.

I changed many settings in the bios, most notable was under-volting the CPU to 1.16V (which it passes 12 hour stability tests just fine).

I believe my issue was simply Prime95 V28 and the stock settings for my motherboard applied far too much voltage than what was required for an already very stressful workload ( I saw in excess of 1.28V). However; Running at 1.16V on an H100i, my chip seems 100% fine now. Never exceeding mid forties in games, and rarely hitting 60 under stress-tests.

Not really sure what was going on there.

[REnso1]

UPDATE: have tested the replacement CPU and it is cooler than the one it replaced.

I replicated the retailers test above at 4x 4.4 GHz turbo, other settings at default and got slightly cooler results with the Noctua http://www.noctua.at/main.php?show=productview&setlng=en&products_id=46 NH-L12.

I ran an OCCT test with the same setup and whereas previously it would overheat and shut down within 2 mins at lower settings necessitating an underclock in earlier tests, this time it completed without problems.

So clearly there was more heat being retained and or generated by the original CPU. I can feel it in the heatsink in fact, more heat is being dumped into the heatsink by the replacement CPU and the HS is getting warmer as a result, suggesting the thermal interface might have been the cause of some of the first CPU's overheating. However there is also much lower steadier power draw with the cooler CPU. A hot CPU will draw more power but a power hungry CPU will get hotter, which is the root of the high transient power draws I observed in the overheating CPU I don't know. The replacement CPU still exceeds http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz Tcase 72.72°C under load with the Noctua.

[REnso1]

OK another update, I have been trying to work out if the heat output of my replacement CPU matches the normal heat output for this model and replicated a bench of the http://www.tomshardware.co.uk/low-profile-heat-sink-mini-itx,review-32826-19.html NH-L12 on Tom's Hardware cooling a 4770K overclocked and undervolted and running Prime95 at 4.0GHz. The results show the same deltaT 47°C between CPU and ambient indicating similar heat output because the deltaT would be greater if the CPU was running hotter with the same cooler. This is quite a neat proof as the figures did match exactly so I am satisfied this CPU is working to spec.

Thanks to all those who contributed.

For those who are having similar problems...

• my conclusion is that the first CPU was overheating as the replacement CPU is running significantly cooler.
  • So this means there are faulty overheating CPUs out there and it is worth exchanging them if you have one, but this is happening in the context of a CPU type which does run hot in bench tests and requires some BIOS adjustment and a better cooler than the Intel retail cooler to bring the temperature closer to spec (Tcase updated to 74.04°C) under heavy load but it still tends to exceed this even under coolers with an appropriate thermal wattage.
• FYI The two key BIOS adjustments I made to enable this cooler were
  • to set the power limiter somewhere between 88 -125 W, currently 110W
  • and to set a negative vcore offset of -0.050v.
  • These restrict turbo to 40x in Prime95 v28.5 but allow it to run steadily and keep the temps close to Tcase.
    • Update, for real world gaming I have set turbo multipliers per active core count to 44x1, 43x2, 42x3, 41x4 which keeps temps below Tcase 74.04°C at all times with the Noctua NH-L12.

EDIT long period stability testing showed undervolting resulted in system instabililty in AIDAx64beta.

For Intel these facts indicate a quality control issue. Further to this, with a non faulty CPU, using the retail cooler results in thermal throttling under load and the CPUs are not able to operate to spec either in terms of operating temperature or clock frequency under loads like Prime95 v28.5 with third party cooling like the NH-L12.

This suggests to me IMHO that the package thermal interface relies on thermal throttling to prevent CPU damage rather than providing a sufficient pathway for peak heat transfer, by design. As a result the CPU cannot achieve its specified capability in bench tests like Prime95 without third party cooling capacity significantly exceeding the rated thermal output of the CPU.

[RKimb]

I'm also having similar issues with a new Intel Core I7-4790K and Gigabyte Z97X-UD3H. Using Boinc on Ubuntu 14.04 with Prime Grid tasks I get 90 - 100 C almost instantly. I've not let it persist and have had to reduce the Boinc CPU time to 45% to keep temperatures down. Even non-prime tasks can cause fairly high temperatures. Even with Boinc set for 45% CPU time I get temperatures over 70C. This is with a NOFAN CR-95C cooler, which is supposed to be adequate. What puzzles me is that the reported temperatures go up and down extremely rapidly. 90C temperatures appear to be dissipated in seconds and drop to 45C in a very short time.

[JBras1]

Also having the same issue here, they gave me a brand new CPU but the problem still remains. I have the exact same issues as described earlier, idle-temps are just fine, but as soon as I start stressing my CPU the temperatures are skyrocketing. It fluctuates in a matter of seconds. Playing BF4 Gives me a max temperature of around 80c, Crysis 3 gave me around 90c and stresstesting it makes it go to about 100c. I have tried the stockcooler and an aftermarket-cooler (Scythe Mugen 4) These temperatures are unacceptable in my opinion, I can hear my CPU-cooler blowing like crazy as soon as I try doing something fancy... I've never had that with any CPU before.

Right now I'm not really sure if it's a CPU or motherboard issue, unfortunately I don't have any other motherboard lying around so I can't stresstest with that. I'm using the Gigabyte Z97X-Gaming 5 motherboard and it's flashed to the latest BIOS. I really hope this problem can be resolved quickly, I seriously have no clue what to do anymore and it really starts to bug me a lot. I can't even use my new machine to it's full potential right now.