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?

[AElib]

Are you trying to make fun of me? Because if so,it doesn't work with me :-)

You are so funny, just look at those test results you posted, that say it all. ;-)

[TGrab1]

Hi Craig,

I personally have not run in to an issue with the Blips causing a stability issue. Also the 7% overclock you are referring to when the chip says 4520ish MHZ is actually referring to the boost. OCCT just sees the stock speed at 4200mhz and anything over that it calculates as an overclock including the boost speed.

So let me explain more about why Robert is concerned with the blips. If you look at Zeox's Vcore graph you will see his voltages are all pretty consistent in fluctuation. When you look at say Mine, Roberts, or your own you will see 3-4 Random drops in the middle of the test. There isn't a change in voltage here or load.. so the random drops are suspicious.

The concern Robert has is that you can have your overclock voltage stable for 98% of the time, but these random blips can come at a time that makes the CPU unstable and crash. So to maximize a stable overclock you need a board that performs like Zeox's is with no random blips significantly dropping the Vcore voltage. Asus has confirmed this is not desired or even correct behavior with Robert, which is why he is trying to collect more information to both provide to asus and to help him made a decision on a new board.

I am by no means an expert in this particular area, but I do understand enough to follow and hopefully explain it better.

[idata]

Hello Craig,

I would be happy to post a stock OCCT test for you if there was some purpose to it, but it will have to wait until I get back home on the 10th June. Why do you want me to do this test? Is it to show that I am getting 'blips' at stock settings? If so there's no need ... OCCT at stock does show the blips as I have tested my boards at everything from 4.2GHz up (with and without turbo boost). The blips are there ... but they are much more easily seen with LLC turned on and speedstep/C-states off as this stabilises the voltage (or Windows to set High Performance rather than disable speedstep and C-states).

To answer your question, I have had no BSODs ever since I have built this system. Did you expect that I would have many? Why? After all I am not running my system outside of reasonable boundaries. But what I have done is to set up my system as a dual-boot, both Windows 10: the first is my working system and the second my test system. When I am experimenting with overclocking etc., I use the test system, so if I should have a failure there's no risk of damage to my working system. I also use Acronis True Image so I can quickly restore either OS quickly and painlessly.

I don't think I seriously threatened to sue Intel ... it would be a pointless and expensive waste of time. Intel has never, to my knowledge, published any data which shows voltages and temperatures on the 7700K at anything above 4.5GHz; and as they have offered to replace or refund my CPU there is no chance whatsoever of winning a case against them. It would be a lot cheaper, quicker and more effective simply to delid and replace the thermal compound.

I hope that ASUS will resolve the issue with the Prime motherboards ... and possibly with your TUF board too. It would be great it they could do this with a BIOS fix. At any rate they are taking the issue seriously and they are now escalating it to R&D (albeit with no promises). If they won't or can't fix it then I will have to look at other options, perhaps from other vendors. You have to remember that these boards and components are very new, so it's not all that surprising to find some glitches - ASUS are a reputable company and they are unlikely to walk away from a faulty product.

As for why I'm concerned about the blips ... TGrable has explained that well. If you are happy at stock settings, that's fine ... I am not, because I run software which is very CPU-intensive and every extra bit of speed I can get saves me time. I have also bought the 7700K as an unlocked CPU and I have purchased the Intel overclocking warranty: so I expect the chip to be capable of more than 4.5GHz. But if the motherboard limits this then I can hardly blame the CPU ... which is why I am trying to make sure that the motherboard is performing correctly.

So ... why don't you help and try out the very simple test we have asked you to do? It won't cost you anything more than an hour or two and it will be very useful both to us and to ASUS (and possibly to you too).

Cheers

Robert

[idata]

Hi ZZX ... I agree that LLC is something to be used with care. On the Prime boards I wouldn't go above 5 as 6 and above boosts the voltage (so if you dial in 1.200 it might give you 1.250, for example). So there's a risk that the ringing on a load change could blow the CPU. But I have to say that I have found an LLC setting of 5 to be very good. On the Prime the Auto setting is really weird ... it isn't off as I thought (at least I don't think so). I'll check it out again when I'm back home next month, but CraigEagleFire's test seems to back this up (it looks to me like he has his board set to Auto). As you can see the voltages are really horrible.

Cheers

Robert

[AElib]

I couldn't agree more with you.

The only thing i have noticed is that you guys mess a lot tweaking your Bios, in my case when i build a Machine i don't tweak that much the Bios, first thing i do is to test the Machine at stock and after that OC using the basic settings = Multiplier - Vcore - Ram speed and Ram Voltage, C-States, Speedstep and Speedshift disabled, test once again the Machine, increase voltage if it's not stable or decrease voltage if the voltage applied was enough to keep it stable.

Note: i don't use XMP or LLC.

[idata]

That's pretty much all I do ... however my system was not stable at settings that it should have been. This is why I looked at LLC and this showed me where the problem appeared to be.

I do certainly use XMP as RAM speed is quite critical to my applications ... I also overclock the cache as that also makes a significant difference (in spite of what one reads ... it's probably application-dependent). But otherwise it's only a question of lowering voltages to reduce heat, if that's needed.

Cheers

Robert

[cmack2]

thanks Robert for all your input so far...

i must admit - i am waiting for someone with CompuTronix's knowledge & technical writing skills - for me to do any BIOS adjustments - a lot of what has been written so far fills me with zero confidence - however, all very interesting...

so Robert, you say my voltages are 'horrible' - i am saying by using ASUS BIOS defaults everything is performing really well & fast, brilliant performance actually - & no BSODs - you are most welcome to submit my latest OCCD test to ASUS - & perhaps you can tell them how to improve their product with all your knowledge...

cheers, craig

[AElib]

Piss me off the fact that the 7700K was designed to run smoothly and without issues on Z270 Motherboards but unfortunately that's not the case.

[ASušt]

No! It was designed to run bad and force you to buy a 150$ cooler and a bucket of expensive fans and a bucket of thermal paste for reapplying that cooler over and over again (hoping that it will help) ...and to make you want to buy another CPU ASAP in hope, that it will work better!

Oh, this whole 7700k thing has just got lame... I'm so close to just close my eyes and to give it up.

I never had so сranky PC ever...my girlfriend is less capricious when she has "those days".

All the previous worked just like a light bulb. I press ON and it's ON and working. Although they were built from mid.grade hardware, not the 400+$ MOBO and 350+$ RAM and all that "PRO/OC/EXTREME/K index" crap that is "not recommended to overclock".

But this one.....first it forces me to use Win10...(that is a piece of a 7700k compared to Win7. Trust me, now I know for sure) then the NVMe drive demands a 50 more $ cooling solution for itself for not blowing up. Then this wonderful spike box.....

What next!?

[AElib]

Just for curiosity i tested my Machine with my Bios settings.

CPU Core/Cache Voltage set to Manual Mode, CPU Core Voltage to 1.20v, Speed shift - off, Speed step - off, C-state off

Here the results:

For 4.6GHz i could use 1.17v but i prefer 1.20v

[idata]

Yes, as you can see, with LLC set to Auto, what happens is that under load the voltage increases. So even though you have the voltage set to 1.20, you are getting 1.217 under load with spikes to 1.232. So clearly Auto is not off: If it was off the voltage should drop under load.

So you would need to be careful when overclocking at higher frequencies as your voltages could be going too high without you being aware of it: that's why using a LLC value of 5 or so seems to me to be a good idea.

But you are still not getting the voltage dips ... which is good.

[AElib]

One more test with the previous settings + LLC 5

I will do a 30 minutes or 1 hour test.

[jmill40]

asus prime z270-a

4.5ghz at manual 1.25 in bios

LLC at 5

Speed shift - off

Speed step - off

C-state off

My problem is when i make XPM for my ram @ 3000MHz, i have my temp Increase about 7°

Temp Ram default

CPU PLL : 0.916v max

Temp Ram XPM @ 3000MHz 

CPU PLL : 1.244v max

The solution to decrease the temp when ram is XPM @3000MHz it's decrease the CPU PLL voltage but the problème it's the asus prime aren't options in bios for CPU PLL OC Voltage, CPU PLL Voltage and PLL Termination Voltage.

Sorry for my english, i speak french

[idata]

Hi ... that's a very interesting test: you aren't getting any voltage dips. But the test is also confusing: your voltages are not dropping under load, as one would expect.

The Prime Z270-A boards seem to be a bit hit and miss!

Regarding your temperatures with XMP on ... it's normal, of course, that the CPU temperature would increase as the memory controller is on the CPU. You should be able to reduce VCCSA and VCCIO and that should help.

Could you run the test for an hour?

Cheers

Robert

[AElib]

He should re- run the test but this time with large Data Set and as you said one hour.

I don't mean to discourage him but a 10 minutes test for me it's not a valid for the simple reason that i could run my CPU at 5.0GHz with 1.28V for 10 minutes and it won't crash but if i leave the test running longer it will crash, that said, 10 minutes tests are not valid.

One more thing, it would be appreciated if

he Post a Vcore graph.

One more thing: why to use XMP when you can set your Ram's parameters Manually :-)

[AElib]

Hi Robert

I have done a test and i would like to hear your opinion.

So, have been doing a few test with LLC, normally for 4.8GHz i need 1.25V in Bios and CPU-z reads 1.264v with constantly spikes up to 1.29V/1.30v, this is without LLC, but here is the thing: with LLC enable for 4.8GHz i need 1.335v and the stable Voltage is 1.312V, in a 20 minute test with LLC it just spiked once, so my question is, which one would you chose between 1.264v with constantly spikes or 1.312V with one spike every now and then?

Cheers.

Albert.

[idata]

Hi Albert,

I would really like to see the Vcore graphs for both tests. But it seems that for both you need the same minimum Vcore as per CPU-Z (which is what you would expect ... LLC isn't going to reduce your voltage / current requirements). The thing is that when you say that LLC is off ... do you know that it really is? When LLC is really off, the voltage should not increase under load ... on the contrary, it should drop. LLC is there for the exact reason to counteract this voltage drop. So on the ASUS boards, LLC set to Auto definitely does not appear to be off.

In my view, and I admit that I may be wrong, an LLC value of 5 is as close as I can get to turning LLC off (or rather, with a setting of 5 I am getting the voltage behaviour that I think we should be getting with LLC off). That is, Vcore is a bit lower than the value set in the BIOS, then under load it drops further ... but it never goes above the value set in the BIOS.

The spikes are normal: they are the ringing that occurs on a sudden load change. OCCT with a large data set really pummels the system with a lot of load changes, unlike Prime95 small FFT which gives a constant heavy load (and so no spikes except at the start and end).

What is confusing me though is that firepanda's tests appear to show no Vcore drop under load (with an LLC setting of 5). The two boards I have and TGrable's board ... and yours too, I think .... do show this voltage drop. So we need to have another look at firepanda's test, I think.

At any rate, for my money I would go for a LLC setting that does not result in a Vcore increase under load: I think that is potentially quite dangerous as you could without realising it end up damaging your CPU. So I would not use an LLC setting of Auto on the ASUS boards.

[BNms]

Hello,

I have upgraded my config 3 days ago (new CM, RAM, CPU, AIO)My problem is that my CPU (i7 7700K OC @ 4.6GHz) rises to 80 ° in full play (BF1)And 60 ° at rest ... the Vcore is at 1.28V (100x46)I use an Enermax Liqmax II 240 to cool it, I do not understand why it heats like that ....Even when it is default @ 4,2GHz Vcore 1,16V (besides I do not understand it oscillates between 800 MHz and 4200 MHz, and the Voltage also, why?) It is at 34 ° then 37 ° then 43 ° then It goes down to 34 ° in idle so on .. What is this edge ***?! http://imgur.com/a/mS04B ">BIOS photos => http://imgur.com/a/mS04B http://imgur.com/a/CMXKX ">Photos of the pump => http://imgur.com/a/CMXKXOS: Windows 10 Pro x64 bitsCM: Asus STRIX Z270F GAMINGCPU: i7 7700K @ 4.6 GHz cooled by Enermax Liqmax II 240GPU: EVGA GeForce GTX 1080 FTW - 8 GB GDDR5XRAM: G.SKILL Ripjaws V 16 GB DDR4 @ 2666 MHzStorage:Samsung 850 EVO 250 GBSeagate Barracuda 2TBWD Caviar Green 500 GBWD Caviar Blue 250 GBAlim: CoolerMaster Silent Pro 80+ 700 WCase: Be Quiet! Silent Base 800 Window / SilverKeyboard: Corsair Gaming K70 RGB Cherry MX Red / Mouse: Logitech G500Headphones / Speakers: SteelSeries H Wireless / BOSE Companion 5Screen: ASUS 23.6 "VN247H

[DA1]

Have the same issue with 7700K.

CPU temp goes over 80'C in gaming, and up to 100'C in LinX stress test with throttling. Stress test in IETU easy warmed CPU up to 100'C but w/o throttling.

No any OC, all stock. I check temp with finger - CPU, heat sink, pipes - all these are warm, not hot. Feels temperature is much lower than 60'C, while IETU shows 99-100'C. I tried to replace cooler paste with no luck. It seems to me that the processor was not so hot before. Heating progressed with time.

Now i return CPU to the seller, and hope to RMA.

UPD: about spikes. In IETU stress test temperature graph painted like a comb. With 100'C on high edges and 70-80 on lower one. While CPU has 100% load. Fan was locked with max RPM in UEFI.

[MWood5]

This was originally opened in mid February 2017, it's now late May 2017.

Has there been any progress in the investigation at all? What is the final word? The product is flawed? A bad manufacturing batch? What's the story here?

I was about to pull the trigger on an i7 7700k for a new build but since finding out about this issue I may have to completely revise by build plans and either go 7600k or AMD.

(does this issue only impact i7 7700k or is it 7600k's also?)

[DA1]

While my 7700K on RMA i bought Kabylake Celeron G3930. Due it weakness - overall CPU usage is pretty big compared to 7700K. But temperature is awesome with 800-900 RPM. Even i run LinX for 10 minutes it still 37-40'C (with drops and raises).