Thermal sensor issue i7-7700k?

[CK7]

I have a brand new build; EVERYTHING NEW. i7-7700k is running at stock speeds. I have the RAM set to XMP for DDR4-2666. Motherboard is Asus Maximus IX Hero Z270.

I have found that the i7-7700k reports a momentary (a second or less) temperature spike +25 > 35 degrees Celsius anytime a program is opened, a webpage is opened, a background app runs etc. The temperature blip cascades through the cores in random order; not the same every time. This causes my heatsink fan to constantly cycle up and down. Temperatures otherwise report as steady, normal increases. Peak temperature under Prime95 blend test is 71 degrees Celsius.

Attempted solutions:

I have re-installed my heatsink and thermal paste with no change.

I have tried to manually set my fan speed in the bios. The only setting that avoids this issue is setting the temperature / fan at a constant (and loud) 80-100%. I've tried PWM and DC mode.

I have found a few user reports elsewhere on the web, all reasoning that it's just the way it is. I don't accept that. Opening a folder or browser should not spike temps +30 degrees. Not only is the fan cycling annoying, it puts undue stress on my fan; possibly shortening its lifespan.

What's the answer, if any? RMA?

[idata]

This test by MSI of 30 7700Ks is a useful benchmark for where our individual CPUs sit on the good/bad spectrum:

Based on this (subject to further testing), I would say that I have a 'potato' as I can only just get to 5GHz. Of course the MSI Good vs Potato test is for CPUs that have not been delidded, so we should be able to get to 5.2GHz or higher with a good delidded sample.

What is quite interesting is that the Potato curve is consistently higher than the Good curve, across the frequency range, so we should be able to tell the quality of our CPU just by seeing how low we can get VCore to at any particular frequency, for a stable test with Cinebench R15 (which may not be as good as Prime95 for stability testing).

The full article, which I guess most of you have read, is here:

http://www.tomshardware.com/reviews/intel-kaby-lake-core-i7-7700k-i7-7700-i5-7600k-i5-7600,4870.html http://www.tomshardware.com/reviews/intel-kaby-lake-core-i7-7700k-i7-7700-i5-7600k-i5-7600,4870.html

Cheers,

Robert

[idata]

I've done some testing, that I'll summarize here.

Setup: I've overclocked to 4.7GHz with the memory and cache at stock. I set LLC to Level 6 (Asus Prime Z270-A) as this seems to give the most stable Vcore using a Manual setting. I had Speedstep disabled, Turbo enabled. Windows Power set to Performance. All other settings at default.

Prime95 Small FFT is stable at Vcore of 1.15v, but Prime95 Large FFT needs 1.23v (will run indefinitely at this setting). Max temperature is 79C at the 1.23v setting, Prime95 Large FFT. Maximum temperature blips of 20C (Windows Apps) with a liquid temperature of 29C.

With memory at XMP (3200MHz), the system is still stable, but max temperature is now 85C on Prime95 Large FFT (6C higher) with same liquid temperature of 33C. Windows temperature blips are now 27C (up 7C).

With Cache at 4.5GHz the system is still stable but Windows temperature blips are now 31C.

Prime95 Mixed gives a Core Max of 77C, IA Cores 78C.

Cinebench is not useful for stress testing IMO as it passes everything that Prime95 No AVX does easily and at 5C less. It's also much too short, so the MSI test that I posted above is pretty iffy, unfortunately. However, as my CPU will pass Prime95 No AVX at 1.15v, on the MSI scale I would say that it is probably a good CPU voltage/stability-wise.

I guess my main conclusion is that for stability testing one really needs to run Prime95 with AVX, Small FFT, Large FFT and Mixed ... for at least 2 hours each. My second conclusion (subject to further testing) is that I wouldn't be able to run my system at anything over 4.9GHz (with an AVX offset of -1 or -2) without delidding.

I'll try tweaking some of the other voltages to see if I they make any difference to the temperatures ... but any other suggestions would be welcome!

Robert

[idata]

Yeah, cinebench isn't really relevant since it takes much less voltage to run cinebench stable at a certain clock than it would take to actually have a truly stable system. Personally, I think occt is the best since most OCs considered "stable" will fail occt within minutes if not seconds. For example a chip that can run 4.8 stable realbench with AVX enabled would likely fail occts non avx test instantly. Temps are also higher, but to me stable is stable and I also want the lowest temps possible under worse case scenarios not just under ideal situations and specific loads.

At this point though with the temps most people are seeing, even good chips capable of hitting high clocks at low voltage still have very little head room due to the thermal spikes these chips are displaying at even stock settings. It's ridiculous that previous gen chips run cooler at equal or higher clocks despite needing to pump much higher voltage through them to obtain said clocks.

And I still wonder why the chips clock up to full speed to accomplish the slightest task. Like I mentioned above, just opening a new tab causes my chip to clock up to 4.5ghz and spikes the temp 30-40c.

[idata]

Thanks ... I've been using OCCT for a couple of days and it is really excellent!

[DS3]

I was going crazy past 2 days about that issue exactly, and then i found this post.

Only difference is that my cpu is i7-6700 (non K).

My cpu temp is constant 25 celcius, but whenever I run a process it jumps to 40-45 for a second and gets back to normal.

At first I tought its related to unstable vcore voltage, so I turned off the C states in bios, and yet even with totally stable vcore voltage I got the temp spikes.

I only started looking into it because my fans went crazy, even when PC is idle the fans go full power for a few seconds whenever a background process starts.

My rig:

i7-6700 (non K)

Asus Z170m-PLUS

Nepton 280L liquid cooler

[JSanc25]

I'm glad this is not a problem that I cause while building my PC, it is my very first build and got very scared when temps where spiking. Sigh, why intel, why?

[idata]

Using OCCT I have come across another issue which is clearly affecting my ability to overclock. This has nothing to do with the CPU and I am only offering it here as it may be of use to some of you. The issue is most likely only with my motherboard (Asus Prime Z270-A), but a similar problem may occur with other motherboards.

I have the voltage mode set to Manual 1.32v (red line) with a load line calibration of 5 (which is mid-way for this board).

At the beginning Vcore is moving from 1.312 to 1.296, with 1.312 being the idle voltage and 1.296 the voltage under load (so a Vdroop of 0.016v).

When OCCT starts the voltage drops to a steady 1.296v, as expected. However, as you can see there are blips down to 1.280v, which have nothing to do with load as the CPU load is at 100%. These blips cause OCCT to fail. To get a stable system I am having to increase the voltage by 0.016v (to get the blips up to 1.296v), which results in a temperature increase of about 8 degrees.

The blips also make it much more difficult for me to overclock as I have to set a voltage, check for the blips, compensate and try again.

[cmack2]

hi Robert,

thanks for your input so far...

as i have an ASUS Z270 mobo (latest BIOS 0906) - i thought i should also input with OCCT feedback for you...

i have decided to run @ UEFI BIOS Load Optimised Defaults - everything runs perfectly @ 4.520 GHz...

70 degree temps @ 31 degree water temp...

hope this helps...

craig

australia

[DFara]

Wow... Is it delidded? Because if not, you must have a 1 in a 1000 chip with this temp at these voltages! Or at least you must be running RAM at non-OC clocks. I'm jealous either way Some crazy discrepancies between these chips, man...

[AElib]

Nah, either way, delidded or not the temps aren't good at all.

[cmack2]

we all know,ZeoxZariX, that you have a custom refrigerated cooling solution...

there are others that actually use liquid nitrogen...

however, those extreme cooling solutions are not 'real-world' for most of us...

(& remember Intel is claiming a thermal maximum operation of 100 degrees C for the 7700K)

[idata]

Hi Craig,

You need to run OCCT for good bit longer ... one hour say. Also, if you look at you temperatures, your base is around 1.25v going to 1.27v, which is quite high for 4.5GHz. You can also see that you are getting a blip down to 1.235v, which is what I found could cause a fault. Because your base voltage is high it hasn't caused your (short) test to fail.

I have a stable 4.7GHz (tested with Prime95AVX, not OCCT, but I will try this) at 1.235v, with a max temperature of 79 with P95 Mixed, and 85 with P95 Small FFT (water temperature 33). So you should be able to lower your voltage quite a bit (I suggest you try a Manual setting with a LLC of 5).

But as I show in the graph below (4.9GHz test), the spikes are awful. In this test the average temperature under OCCT load is around 80, but the spikes go to 93. And as you can see, at 'Idle', the spikes are up to 35C (from about 35 to 70). Because of the voltage and temperature spikes (not to mention the high average temperatures), 5.0GHz is not really an option for me. So the best I can get is a 700Mhz boost from 4.2GHz ... while I would have expected 1000MHz.

[AElib]

If you look at you temperatures, your base is around 1.25v going to 1.27v, which is quite high for 4.5GHz.

I fully agree with that, at least in my case i need 1.15v for 4.5GHz and i have a Z170 Board.

Craigeaglefire

remember Intel is claiming a thermal maximum operation of 100 degrees C for the 7700K)

They Claim 100c for the 7700K but i don't believe that much.

[cmack2]

hi Robert,

all good info - just ran the ASUS default stock clock test for a bit longer as you suggested...

& yes, i could probably have slightly lower VCORE - but i am not actually sure what that would actually really achieve...

as TGrable says - the 7700K is a beast in performance - i have yet to learn all the nuances of overclocking & learning settings from you guys - but 4K gaming, video production & encoding & general computer functions all seem to be working very well & stably - plus reasonably good actual benchmark results...

cheers

[idata]

Well at least you should be pretty confident that you have a stable system ... But you can see that your Vcore dropped down as far as 1.22v, which is getting close to the min voltage for a stable 4.5GHz.

By changing to Manual and a lower voltage, LLC not at Auto (which really means off), you should get lower temperatures and also less down-blips. You can also set IA AC/DC Load Line to 0.01 (http://edgeup.asus.com/2017/01/31/kaby-lake-overclocking-guide/3/ The Kaby Lake overclocking guide - Page 3 of 5 - Edge Up ): according to Asus this helps to keep Vcore nearer your CPU voltage setting in the BIOS. But if you're not interested in going above 4.5GHz you probably won't see much benefit, except less fan noise perhaps. On the other hand you should be able to get up to 4.8GHz with little effort.

Cheers

Robert

[AElib]

which is getting close to the min voltage for a stable 4.5GHz.

Since when 1.22v it is close to 1.15v?

[idata]

True . Well, at least it's closer than 1.27.

[GMaci]

Hi, I'm having same issues with a brand new pc. Hoping in fast reply. Thanks a lot

[vbro]

disable the BCLK voltage adjustment in the motherboard bios. That will reduce the natural overvolting and should reduce the temperature of the spikes but, won't get rid of them completely

[RonaldM_Intel · Accepted solution]

Hello Everyone,

We appreciate the feedback you have provided, and your patience as we investigated this behavior. The reported behavior of the 7th Generation Intel® Core™ i7-7700K Processor, showing momentary temperature changes from the idle temperature, is normal while completing a task (like opening a browser or an application or a program).

In our internal investigation, we did not observe temperature variation outside of the expected behavior and recommended specifications. For processor specifications, please refer to the https://ark.intel.com/products/97129/Intel-Core-i7-7700K-Processor-8M-Cache-up-to-4_50-GHz Intel® Core™ i7-7700K Processor Product Specifications.

Most motherboard manufacturers offer customizable fan speed control settings that may allow for smoother transition of fan revolutions per minute (rpm). Please consult your motherboard manufacturer's manual or website for instructions on how to change default fan speed control settings.

We do not recommend running outside the processor specifications, such as by exceeding processor frequency or voltage specifications, or removing of the integrated heat spreader (sometimes called "de-lidding"). These actions will void the processor warranty.

Kindest Regards,

Ronald M.

[WMeed]

3 months waiting for Intel to come out with a solution , and now this ?! This is all you can say ?! We know already what you've just said .. You know what never mind this would be my last product from Intel .