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.

[HFigu1]

I purchased the ASROCK Extreme 6 paired with my i7 47790k. Here are the results of the new system.

[idata]

jfig04-

I have the same motherboard as you, but get much higher temps. Are you using stock cooler, or something else. Let me know, as I'll get whatever you have. Your temps appear great!

Got a batch number just for future reference?

What is your bios version?

[HFigu1]

TAS3086-

Bios version is : P1.30 (1.40 is out. Everything is working fine so I wont update it.)

Cooler : Corsair H110

Case : Fractal Design Arc XL (just in case you were wondering)

Processor Batch: X437B319

[HDi_P]

Hi all!

I'm fighting with the same issue: system is i7 4970K, batch L420B916, CPU spiked to 100 when using stress tools such as IntelBurnInTest.

I've downloaded XTU a used Ken Settings, now TDP is more or less in check (i expect short TDP spikes in Turbo), reaching 89ºC maximum temp with BurnInTest.

Here's XTUs default settings and the proposed ones (motherboard is Asus Z97-P with latest BIOS):

I didn't clear the CMOS yet, but I would like to show the current situation, and report enhancements if I can.

The following are the results of 2 min of Stress testing in XTU.

Approximately, maximum achieved values were:

CPU Temp: 87ºC

TDP: 89W

CPU Frequency: 4.15 - 4.2GHz

CPU Core Voltage: 1.16V

Any thoughts?

Im running the stock cooler. Also I have a Coolermaster TX3 from an old i7-870 system, but from what I've seen, I does not provide a much better solution.

[HDi_P]

I've found that using X.M.P Profile settings on my 2400 DDR makes idle at 13W according to CoreTemp. Without XMP, 4W and idle cores at 30º-35ºC which is good.

Instead of XMP profile, I choose to set the memory speed manually and leave the remaining settings untouched.

In anycase, high temperatures under load still occur (between 75 and 85ºC).

I don't know what to do! This is a CPU fault Intel?

[KenF_Intel]

Hi IndioColifa,

Can you do a test with memory running it's stock speed? No XMP and no manual memory overclocking?

Thanks.

[HDi_P]

Hi Ken! thank you for your response.

First I undervolted the chip 70mV to push temps down a bit and mantain 1200MHz (2400DDR) memory rate. After I checked this is stable, I finally removed the stock cooler and retried the Coolermaster TX3 installed in my old i7-870.

To my surprise, i managed to achieve better temperatures, even with XMP enabled.

Here is XTU stress with MCE (all cores up to 44x) and no-MCE. Without MCE, stress temps average 75ºC. It's not perfect but much better than with the stock cooler.

I must also re check what X.M.P does apart from elevating voltages, because with XMP the boot screen says "4027MHz instead of 4000MHz" for the processor.

So, effectively the cooler made the most difference.

I will re-do testing with all default settings again, 1333MHz DDR3, and the TX3 installed.

Thank you.

[YKar]

Ahhh, problem solved when I changed the CPU cooler to Cooler Master Hyper 212x! Here's my new temperature and I am very satisfied with it!

This is when it's idling

This is when running games

This is when running Prime95 for 15 minutes

I can say switching to third party cooler is amazing!

[HDi_P]

Well I played a bit more with my 4790k setup and achieved lower temps.

Here are my findings; some maybe correct or apply in other users configurations.

• First, cooler reseat. I re installed my Hyper TX3 cooler checking if thermal paste was distributed correctly over the chip. It was bad, I tried to "tint" the heatpipes instead of the traditional "dot" application: https://onedrive.live.com/redir?resid=B8273AED3A35C0A4!8941&authkey=!AH7EJuRcJbPDASs&v=3&ithint=photo%2cjpg Microsoft OneDrive - Access files anywhere. Create docs with free Office Online.
• Correct application (using Coolermaster thermal compound): https://onedrive.live.com/redir?resid=B8273AED3A35C0A4!8942&authkey=!AO6hTKYsLpiJbBE&v=3&ithint=photo%2cjpg Microsoft OneDrive - Access files anywhere. Create docs with free Office Online.
• This lowered my idle temps 5°C and load temps 10-15C° in comparison to the stock cooler and in comparison with a big heatsink but without correct thermal compound application. So first check your thermal compound. I also got PTI G3606, but I think it'll be marginally better, so I didn't try.
• I reset the BIOS to defaults and run a good ol' IntelBurnInTest at Maximum Stress level. I got:

DEFAULTTX3+ fix thermal compoundMinMaxMinMaxMinVID0.692V1.210V0.692V1.264VCore030943081Core127932581Core228932681Core328882578Package34943082Package W4.37W96.76W3.92W83.32WIA W0.09W91.38W0.09W79.07WUncore W5.27W9.95W4.71W9.86W

Good improvement (and this test in particular pushes this CPU more or less like Prime95 with AVX).

• My first conclusion was that this chip IS HOT , and the stock cooler is not enough for the 88W TDP at high loads. A decent cooler plus decent thermal compound may keep the CPU operating at near it's rated TDP, a bit less or a bit more. I'm not totally sure since other users reported crazy temps with other cooling systems, much more powerful than my good ol' TX3.
• I researched a bit about X.M.P, and why it generates higher temps (in order of +10°C): at least in my case, the DDR2400 kit profile specifies a DRAM voltage of 1.65V and a Memory Controller voltage of 1.25V! This is VCCIO and/or VCCSA in Haswell platforms and seems to add more heat generation.
• Since I got XMP enabled, because the Haswell chip scales memory speed very well giving appreciable bandwidth gain at 2400MHz, I looked to lower the temps with XMP enabled, both in load and idle.
• I undervolted my chip and run AIDA for 3 hours and seemed stable, for now. I used Offset Setting with -0.06V. Note that Adaptive setting will apply offset only in Turbo Mode, not at idle. So to generate lower idle power use, Offset setting should be used.
• I found that Dynamic Storage Accelerator enabled generates higher idle speeds when enabled (I dont know why yet). 13W vs 5W at idle.
• ASUS Z97P mobo settings for Turbo Mode thermal power limits are "unlocked" (4095W for Long /Short Duration) and Unlimited current. Since XMP pushes the processor exceeding the 88W TDP at stress loads, I wanted to setup power limits 88W/110W short duration/105A current Max (I know this would limit my maximum Turbo speed but I dont want to bring down this juicy 2400MHz memory bandwidth). Unfortunately, seems that ASUS BIOS contains some kind of bug since the reported limits in AIDA are completely wrong when i enter them in BIOS (e.g: 88W Short and AIDA says 400W). I already contacted ASUS technical support.
• I'm pretty happy at my current rig operation parameters. This chip is a BEAST, exceeded my performance expectations, but cmon Intel: THIS CHIP IS HOT. Next time, skip the stock cooler and include a card with a warning: "DEAR INTEL CUSTOMER, BUY A GOOD AFTERMARKET COOLER WITH THIS CHIP. THIS THING IS HOT". Don't play with customers, Intel.

[KKowa2]

Just got my sample, S-spec: SR219, Batch: L434C225.

This is fresh BIOS on default settings and stock fan.

Everything is stock, nothing was overclocked, even X.M.P. is disabled.

And here it is. 50 degrees Celcius while idle in BIOS. WOW. I mean WOW...

RMA request already posted.

Chris

Edit:

I manually put it at 4.0GHZ with 1.000V and tested it in OCCT:

[JMahr]

System : MSI Z87-GD45 Gaming Bios 1.90

Intel i7-4790K

GSkill DDR3-2400 TridentX 1.65 4x4GB

EVGA GTX780 ti

Sammsung 840 EVO mSata

SeaSonic Platinum 1000W Power Supply

Coolermaster Hyper 212 evo

Temperatures in Intel Burn Test spike immediately to 100C. They are about 29C to 30C at idle. I just finished reading this thread and will try the suggestions that Ken gave. Just wanted to know if anyone with my same mobo had these problems and what was done to correct it.

[KenF_Intel]

Hi Joshwaa,

Hopefully the recommendations will work.

Please let us know the results, especially if you need to tweak the settings to get good results.

I've created a document that contains the information from my earlier post as well as some of the settings that have been reported by other users.

That document is located here:

[DVu3]

Hi all,

After gone through a whole set of mobo, cpu and ram, I gave up and returned the entire system back to the store. I got the same problem with overheating, too! But instead of running stress test, I worked with a program called Gaussian (computational chemistry) and every single time when it's working on the job, the temp was shooting up to 100C in just a couple seconds, then it throttled down. I used either 4 or 8 threads while running Gaussian jobs (Hyper Threading seems quickening the jobs by 1-2% so why not, right?). I tried almost everything to get the temp down but I failed. The first combo was i7 4790K with ASRock Extreme4 and 16GB of RAM. This combo is history now.

I just got a whole new setup of ASUS Z97-AR with another i7 4790K and 16GB and the same thing happened again. 100C within just a couple seconds of starting the calculation. I tried the setting of ken-Intel but it still creeps up to 100C. Turning Turbo Boost off seems to help with the overheating, full 100% CPU loading at 88C using stock cooler. Fan is set at 100% of all time and room temp is 18C (winter time). Waiting for the Hyper 212 Evo though to see if it's gonna help.

One thing though, I use a minimal Ubuntu version (no GUI, just terminal) to host the Gaussian program, I don't know if it's gonna be sufficient enough to monitor the performance of the CPU. Checking the MHz of the CPU always list at 4001.00 MHz while Turbo Boost on, and 4000.00 MHz without Turbo Boost. Even though Ubuntu kept sending CPU throttling messages, grep cpuinfo always show 4001.00 MHz so I had no clue if throttling ever worked. Probably it did cause temps never passed 100C (just 99-100C all the time).

I have another i7 930 machine running Gaussian also with stock Intel cooler and it never passed 70C at all, even though its TDP is 130W. What's up with that?

Any idea on how to fix this overheating issue?

[REnso1]

RigItPC I think you may need a better heatsink, sounds like Gaussian puts a lot of work the CPUs way and assuming the CPU is a good one the temps you describe suggest it exceeds the workload of stress tests like Intel eXtreme Tuning Utility and AIDAx64 even, approaching recent versions of Prime95 which provides the highest test bench CPU load of any of the tests I have used by a fair margin.

The CPUs do vary and I summarised my own findings /message/253580# 253580 here. I likewise had to replace one and that improved matters.

I am using a Noctua NH-L12 heatsink and its quite a bit better than Intel stock and even so it is not sufficient with the GA Z97MX Gaming 5 which seems to run Haswell I7 CPUs even hotter than expected. Two of these mobos were RMAd for heating by SCAN tested on I7 4770K and 4790K. So some overheating is mobo related and some is CPU.

With this heatsink I can run AIDA64 with turbo disabled for ten hours straight maxing at 69°C (stable using default with turbo off, medium LLC and XMP). But I cannot enable turbo without applying watts and amps limiters which restrict 4x turbo below default and max temps reach 73°C. I dont want to enable turbo with limiters under these circumstances because I suspect local core temps will rise too high in 1x and 2x turbo modes with limiters which work for 4x mode, since all that power, instead of being divided by 4 can be poured into one core without restriction, if I understand correctly.

I would suggest testing your CPU with the XTU using advice in /docs/DOC-23517 Kens troubleshooting summary if you are able to run it in Ubuntu and compare like for like to see if you have a hardware anomaly and if not you could try getting the best heatsink you can fit in your case, better heatsinks definitely do reduce the temps under load with this CPU.

[JMahr]

I did the adjustments that Ken gave us. Those allowed the computer to run but when I tried to reboot the computer would blue screen when coming back into windows. I eventually found out that setting the Vcore to 1.100v was the issue. Had to leave it on Default/Auto. So far it is working ok.. I have an CoolerMaster Hyper 212 Evo and still the processor gets to 84C during the Intel extreme tuning stress test. If I run the memory stress test in Intel Extreme Tuning the temp goes up to 89C. I have put the cooler on with Antec Formula 7 and the temps where about 2C more and when I had the Noctura TIm it was about the same 2C more. Now running with Coollabratories Liquid metal pad and getting the temps I stated above. I know the the origonal 100C degree temps were the fault of the Motherboard so I can not blame Intel for that. But considering I upgraded to this from a i7-4770K that would run all cores at x42 multi and turbo 1 core at 45x multi and never go above 75C with the same mobo, ram, and cooler I am extremely disappointed and feel that I truly wasted 300.00. In real world comparison I got a CPU that is supposed to be a step up but was actually a step down. I do not know if I just got a bad cpu or if this is how it is supposed to perform.

Other thing I forgot to mention. In the BIOS the 4790K idles at 41C where the 4770K was at 30C, also at idle in windows the 4790K is at 32C where the 4770K was 29C.

[KenF_Intel]

Joshwaa,

A couple of questions:

1. What does XTU show as the CPU Core Voltage while the test is running?

2. What does XTU show as the CPU Total TDP while the test is running?

Please post a screenshot of XTU if possible.

[JMahr]

Keep in mind that is with a CoolerMaster 212 Evo with a Corsair SP 120 High Performance PWM running at 100% fan speed. In an open test bench. Intel burn test on High causes Current Limiting almost immediately. This is the 21st Intel processor I have purchased for personal use (first build 386SX 16Mhz) and the first one I have ever been disappointed with. I know you guys make excellent products maybe I got a dud. The other sad point is the K is worthless I will not be able to even play around with overclocking this one.

[YNomi1]

I have been watching this thread from start. I don't want to set fires or something, but i believe that all the chit-chat here is POINTLESS. I watch Ken-Intel (which i honestly thank for all the help trying to provide) trying to answer a problem that NEVER should have been there in the first place. Everybody else is trying to follow Ken-intel's instructions, setting caps and limiters to everything so as to limit the overheating that the BEST HASWELL K CPU is producing. This thread should be about how to overclock 4790K and not underclock it. PLEASE INTEL just admit that you screwed up on this one. I am also using intel cpus from 8088 4,77Mhz without ever having problems, but with this one, complete disappointment. Please intel do the right thing and offer a refund. That's the least you can do to everybody that coughed up 330 USD of their hard earned money, for a defective product.

I know that my comment is not a constructive approach to this thread, but in my life, i am used to tell things the way they are. And because of my 30 year experience in computers, the only term with which i can describe the situation is the CharlieFoxtrot acronym.

[ak25]

I have a similar problem and I've not been able to get some clear answers. I have an i7-4790K with stock intel fan. If I under volt vcore -0.050 volts temp with mprime is 85-86C tdp max is 82 watt (according to intel tool).

-

This is the 11th lga intel system I've built and i'm 85% sure the fan is installed correctly (I can see the pins on the other side of the motherboard and it seems firm (i actually prefer amd's fan attachment as it is more positive - but this is a side issue).

-

The voltage readings are: cpu 1.71 and vcore (with my undervolt) 1.071 (1.12 before undervolt but then I hit 100c).

-

What is not clear to me (and I would like intel to respond if possible):

Is it safe to run the cpu for a long time at 85-88C ?

Is the cpu voltage of 1.71 correct

Do I have a faulty cpu (is it worth warranty exchange) or is this the expected behavior with stock fan ?

-

I did contact intel and they are offering a warranty exchange but it is not clear to me if the hassle of replacing the cpu will improve my situation. It is also not clear to me if it is ok to run the cpu at 88C for 24 hours.

-

I don't totally object to the under volt because the system is stable and I sitll reach turbo speed and I had not planned to overlock (I have a K processor because during blackfriday it was less expensive than the non K )

-

From this thread it sounds like 88C is actually pretty darn good but it is not clear to me if folks who are seeing 100C are running at higher tdp than i am.

Oh yea I ran the intel diagnostic tool and the processor passed.

[ak25]

Here is a copy of the xtu output from a stress test today with mb default vcore. Please let me know if this looks ok and if it is safe to run the cpu for long periods at this temp or if the cpu is defective.

[CPapa1]

Like other people, I've got an i7-4790k that ran at disturbingly high temperatures. The HSF is a Coolermaster Hyper 212+. It's not the latest and greatest solution, but certainly better than stock. My BIOS configuration was not set to overclock (I thought).

I'm running Linux Mint 17.1. When I would run prime95 with small FFT, the temperature *immediately* shot up to the high 90s and stayed there. Thanks to posts in this thread, I was clued in that XMP might be the cause. Sure enough, I had it enabled. Disabling XMP resulted in small FFT temperatures in the low 70s instead of the high 90s.

The only reason I enabled XMP was because I have PC2133 DDR3 memory that the BIOS wants to run at PC1600 by default. I thought the purpose of XMP was to use the correct memory timings specific to the DRAM modules? Why the heck is it also affecting CPU power consumption??