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?

[PYoon1]

Can someone from Intel please provide an update?

[NDe_S1]

Intel,

Please give us an update on this issue!!!

[ASušt]

Have registred to add myself to the "spiky heat" club.

Any link clicked, any programm opened just a little increase in CPU load instantly causes spikes.

Temp jumps 30-> 45/60 and drops after 1 sec.

Fans howl all day long.

Well nothing new, you know the story.

Pity that I can't return the cpu =( Would give it try to replace it with another 7700k if it is possible.

Hope Intel will figure something out. As for now - constant coolers' speed is the way to go for me.

P.S If someone is intrested:

I7 7700k spiky lake

Asus IX Formula

Corsair Vengeance XLP 3600mhz

Noctua u14s CPU cooler.

[TGrab1]

I think the spiking isn't anything alarming in and of itself. All cpus will spike when usage goes up. That is what happens. Them higher the frequency or state the more Volts being pulled and the higher the temps generated.

It is just more more noticable on this platform for most of the reason stated. Also keep in mind this is a 14nm chipset. That also leads to less actual contact area between various parts of the DIE/IHS/Block. Then you throw in TIM and Spacing/gap issues and it can even get worse.

How many people are running a factory system with one of these chips and noticing a fan noise issue? My guess is not many and that is because those system should be configured to not instantly ramp up over the spikes. So a good portion of the fan complaints can be sorted by configuring your fan profiles normally in bios to not instantly ramp up or down on quick temp changes.

I'm not saying the 7000 series isn't without its problems. I am just saying that the temp spikes aren't really that alarming as you will notice that on most cpus. Even a few AMD systems I have laying around see spikes when opening or closing browsers, windows services running, etc.

All I can suggest is just making sure your temps are in a safe spot and don't push your temps higher than is safe for the sole purpose of getting an Overclock. Not without upgrading thermal solutions first I mean. Most overclocks are solely for bragging rights at this point anyways. The real world difference in games and general usage is very minimal. The only time you will see decent gains is for benchmarks and the like.

So if you are having heat issues just run it at stock. You are not missing out on anything at this point in time.

[idata]

I think that a temp spike from 30c to 69c at stock when idling or opening a browser is nothing close to normal, even Intel admits that its not normal. I cant run prime95 for more than 5 minutes before the cpu hits 85c. If u don't have a 7700k or the spiking issue I dont understand why you are even here as this thread is for people with the issue or a solution for it.

[ASušt]

to TGrable

Your thoughts are clear and logical.

I agree when the CPU is hot under a benchmark/game/demanding task.

But seriously.

Surfing through Intel's forum that doesn't even contain pictures, makes spikes up to 65c (from 30c).

If i click a link at least every 3 sec., it looks like CPU is constantly at 65-70c.....with "forum load".

If I open a folder, open a picture, launch notepad/paint etc.

I don't believe that the CPU under a huge cooler can actualy generate that much heat in 2 sec.

And, most importantly - cool down from 65 to 30 in 1-2sec.

I tried a stress test. After 1 sec after start, the CPU is already 65c. And it stays 65c for 1 hour. After stop - cools down to 30 in 10 sec.

We don't have heat issue, we have spike issues.

P.S to Zaphoo,

Let me just shake your hand.

[DBux]

Please man... A LOT of people has the 'spikes issue' here at stock, it's the origin of this thread. Now it's, in a wrong way, becoming an overclock thread, but this is not the key. You must read all messages to understand that, of course, exists a real issue with NON OVERCLOCKED 7700k cpu temps, going from 30º to 70º only opening browser (as mine), or reaching more than 90º when gaming (of course, with a good cooling system). Then you must not think we are claiming nonsense like newbies and don't try to take importance from this, in my opinion, Intel BIG mistake. We are still waiting for a solution from Intel (or, at least, for a reasonable explanation).

[TGrab1]

In response to you, Zaphoo and Links.

First off I have been here since this thread basically started and if you go back you will see I have made numerous posts a long the way.

I have gone over what I "think" the issue is. I am saying that the issue of the spikes isn't where your focus should be. Yes the cpu will spike up and down from various tasks, but the spikes will happen on any CPU that has variable states and frequency/volt ranges it can pull between active/idle/sleep/parked etc.

The spikes are only a symptom of the true issue which is the heat this chip produces and the corners intel cut on their TIM solution.

People are mentioning high temps during a stress test, or high temps in a game. Those are TEMP issues and not part of the spike. The fact the temps spike from idle to 65+ in under a second from a slight load is ALSO a temp related issue. The spikes are not a problem with the actual CPU. They are not a problem of the architecture or anything that is new in that spectrum with kabylake.

So im going to do a little ASCII demonstration of what I have been trying to say the problem is.

------ will be your die and o will be a good contact. / or \ will be no contact in a sloping pattern and a space would be just no tim coverage.

So in an Ideal world your TIM or Soldered Solution via my model would look like this

0000000000

---------------- This would be a full coverage with no issues.

The problem with Kabylake is first they are using a CHEAP TIM with less thermal conductivity than most would like. Second the TIM application is not great on the few delids I have personally seen (my own included). The next problem is with how an uneven silicon application (the adhesive between Chip and IHS) leads to bad contact.

So first is air bubbles in the TIM application

00 0 00 00

------------- The gaps are areas where there is no good thermal exchange and heat just builds up as Air is a bad conductor of heat.

\\000///

--------- This would represent a silicon bead that isn't consistent leading to poor or even no contact between the DIE and IHS. Maybe there is just a thicker layer of TIM in there that helps bridge the gap, but the more TIM the less thermal Conductivity between die and IHS. So once again you have an area where temps are higher because there is bad contact.

On a older chip these connection were either soldered (which would be very similar to what you see with people using liquid metal on a delidded chip) or they manufacturing process was not as small.

000 00

On a 14 nm Process You would have -------- So having a tap on a already small surface area really hurts your ability to transfer the heat from one source to another.

0000 0000

On a 22nm process you would have ------------- So even though your coverage is not perfect you still have plenty of surface area making contact that the issue isn't as bad or noticeable.

So all the problems you are seeing with high stress temps, huge spikes, massive jumps from idle to use all stem from the one underlying issue.

This issue is that the way intel handled the contact between the DIE and IHS are inconsistant. This includes TIM and Silicon bead application. This is also why some chips run much better than others on temps. Being only 14nm there are very few margins for errors on contact between DIE and IHS. This is why Delidding on these CPU's are seeing such massive gains and changes in thermal performance.

So in the end you have

Perfect stock chips - good temps at various states

Average chips - small TIM or Silicon Issues you will see higher temps and spikes, but are overall safe.

Poor Chips - Bad TIM application and Bad Silicon consistancy leading to terrible temps and massive temp spikes.

Even if you consider all areas even that is still 2/3 of the chips with issues. I can tell you now that from my experience those perfect chips are probably less than 10-12% of the ones on the market. So most people on this forum will be in the bottom two categories.

So once again, if you are seeing temps under 80c the spikes are not that alarming and are part of the bigger underlying issue. If you are over 80c check there other things you CAN control first. Then you will need to contact Intel for an RMA or risk delidding your cpu.

One last time though I will say the SPIKES(most cpus spike when changing states) are not the concerning thing. It is the other underlying issues that are to blame for the spikes that need to be addressed. Which are what we should be focusing on as a whole. The Intel Reps are already aware of this i am 99% sure. This issue has happened several times in Ivy and Haswell. The issue got much better later on when they switched to a better TIM. I suspect we will see a very similar outcome here as well, but unfortunately it doesn't help those of us who already have a 7700k with issues.

[ASušt]

Not saying you are wrong,

But I would like to hear from Intel.

If my fans didn't ramp up their rpms I probably wouldn't even notice spikes.

There is no setting, that would allow to set a delay after which that would happen.

As for deliding, it is a procedure that woids the warranty 120%, and you need to have some experience to do that, so I guess this can't be counted as a general solution.

And for the conclution: I'm here not to show how smart I am and how stupid are you. I'm here in search of a solution or at least an advice.

....oh gods of chaos, make Intel answer already....

[idata]

That all sound entirely reasonable, and since delidding has been shown to pretty well fix the issue I think it's not unreasonable to conclude that this is the fundamental problem. The fact that the 7700K runs at a high clock, has AVX support, is 14nm etc., only makes it worse.

Here is what my CPU looks like. The spikes occur when an app like a browser is opened, but they also occur when the system is idling.

I would like a clear statement from Intel saying either that this is entirely normal ... or that there is indeed a problem (as we know there is) and that they will fix it. After all, if a car engine was overheating in this way then the car manufacturers would do something about it, for the simple reason that they would not be allowed by the public or the press not to. In this case the issue is rather contained to this sort of forum, and Intel may feel that if they lie low that we will go away. We should not do that ... instead we should escalate the issue until Intel are forced to act. It would be interesting to hear some suggestions on how we can do that.

Cheers

Robert

[idata]

Wow! Looks like the 7700k has more problems than the 6700k.

I know I don't own the 7700k, but the 6700k I have experiences similar "spikes". Luckily I'm using a water cooler to keep the temps down.

Surely the temperatures should gradually increase rather than jump straight from cool to hot?

I wonder if the processors are drawing too much voltage on the default BIOS settings, like mine was? Intel, please investigate.

[TGrab1]

So I keep hearing recall thrown around and compared to a car.

When it comes to a recall there are lots of factors that come in to play.

Cost of letting the defective products stay.

Effect on image of the company.

Cost of any failure to recall penalties

Type of risk involved.

If the cost of the recall is more than those things, then they will not do the recall.

Which leads me to my conclusion that I don't think they will do a recall. They will probably allow RMA's in most circumstances, but the amount of people coming forward with issues and those that even know they are affected is a very very small percentage compared to those that actually have them.

So the cost for letting them stay... is just the cost to replace a chip or let it be RMA'd if it does fail.

Effect on the company image is negligible since few that own the chips are aware of the issue and this is also something they have done before without a concern.

Not a big enough or dangerous enough issue for them to worry about penalties and even if they did.. they would also probably be negligible.

Risk is just a dead CPU. Maybe some data loss or loss of important information, but generally they would not be help liable for these things.

So In a cost/risk assessment I don't see them ever choosing a recall in this situation.

I do, however see them offering RMA to those unhappy. I was offered one when I complained about temps on mine, but decided I would delid as the issue would probably persist after unless i go a golden chip.

Anyways that is my assessment so take it with a grain of salt.

[DBux]

@TGrable, I'm with Links. I understand your point, and I'm sure it's true in lot of cases. But when I bought my i7700K nobody asked me if I want a 'perfect, average or poor' CPU (of course it must be perfect!), and this is why INTEL can't avoid it's responsability.

But returning to the main issue, perhaps you don't remember the important info that some people shared. I mean that seemed physically improbable, for me, to go from 30º from 75º and return to 35º in two seconds; this gave me the clue (other people shared full test about confirming) that the problem isn't perhaps with real temps, but with sensor readings. I know that CPU temp is a CPU internal value, but I never noticed externally this temps change except for the fan going up and down continuously (no heat in case, no heat in cooler, no heat also touching with finger nearest possible CPU, etc.). I mean that your explanation was very reasonable, but little useful for a lot of us, I think.

Now I'm at that point of despair that, if I knew an option to delay the CPU fan reaction to the temp CPU sensor one or two seconds, could be my solution...

p.s.: Ah, and yes, anyway I also changed my Gigabyte bios settings because a lot of people was right, it was giving too much voltage to CPU (I'm still at stock, no overclock). Now I'm not reaching 90sº or more, but 80sº...

[TGrab1]

DanBux wrote:

@TGrable, I'm with Links. I understand your point, and I'm sure it's true in lot of cases. But when I bought my i7700K nobody asked me if I want a 'perfect, average or poor' CPU (of course it must be perfect!), and this is why INTEL can't avoid it's responsability.

But returning to the main issue, perhaps you don't remember the important info that some people shared. I mean that seemed physically improbable, for me, to go from 30º from 75º and return to 35º in two seconds; this gave me the clue (other people shared full test about confirming) that the problem isn't perhaps with real temps, but with sensor readings. I know that CPU temp is a CPU internal value, but I never noticed externally this temps change except for the fan going up and down continuously (no heat in case, no heat in cooler, no heat also touching with finger nearest possible CPU, etc.). I mean that your explanation was very reasonable, but little useful for a lot of us, I think.

Now I'm at that point of despair that, if I knew an option to delay the CPU fan reaction to the temp CPU sensor one or two seconds, could be my solution...

p.s.: Ah, and yes, anyway I also changed my Gigabyte bios settings because a lot of people was right, it was giving too much voltage to CPU (I'm still at stock, no overclock). Now I'm not reaching 90sº or more, but 80sº...

Just out of curiosity what cooler are you using?

Also It is possible for the temps to go up and down that fast sadly. It is because the cpu cores are heating up faster than the heat can transfer to the IHS then into your cooling solution. Which as I have stated many times is a TIM contact issue. I was having those EXACT problems before I delidded (which I do not suggest doing for most people).

When I say perfect CPU I am referring to the Thermal interface efficiency and defects in application etc. All of these chips are "perfect" in the sense they pass intels quality testing before getting shipped. These chips are within their operating temps and such, but they do run hotter than most people like. Which is do to the tim quality :-(.

I would recommend that you email Intel and ask for an RMA. If you explain your issue and temps you are seeing I am sure they will authorize one.

I would still be curious to what you are using Cooler wise though :-D.

[DBux]

I'm using liquid cooling Corsair H100i, the best I could get at that moment. I think it must be quite enough!

[TGrab1]

Ya 240mm AIO should be plenty.

The good thing though (as much as corsair link can be a turd) is that corsair link should allow you to setup a fan profile that keeps your machine from ramping up so often.

Also since I am curious did your cooler come with the TIM already on it or did you have to apply it yourself? If you did it yourself what application methodology did you use?

Not trying to place blame on you, just curious if there is anything else I might be able to suggest to increase your thermal efficiency a little more.

I mean who am I kidding I tried tons of things when I first got my CPU and noticed the temps I was getting on custom water. I think I was pulling 72-75c on stress tests at stock settings. Now on those same tests I pull high 40's at stock. It is stupid how bad the TIM on these is >.<

[idata]

I've got the H50 or H55, I did apply the thermal paste myself, twice. Added a little more on the second try in case i didnt add enough on the first try, although it looked perfect when i removed the block of the cpu

I've tried changing every settings suggested in the BIOS to make the spiking go away, I even set the vcore so low that windows wasnt able to boot. Since I don't have a lot of experience with BIOS im not sure if i did it correctly. Still at the lowest vcore where i was able to boot i had the same symptoms and temperatures.

I checked HWMonitor just now, and after a day of only gaming my top temps are:

Core # 0 70ºC

Core # 1 70ºC

Core # 2 68ºC

Core # 3 64ºC

MAX VCORE: 1,264 V

Since its under a month since i bought the CPU, and the website i bought it from has a 45 day return policy im gonna ask them to send me a new one and hope i win the lottery. As long as it's an manufacturing defect i can by law request that the seller repair or replace the faulty chip.

And I'm sorry for "going after you" earlier TGrable, that wasn't fair of me.

[TGrab1]

No worries I didn't really feel attacked or anything.

I will say that the TDP of the H50/H55 is probably close to what you are pushing. So you won't have a lot of thermal overhead with that cooler.

Also did you try the X pattern application method with your paste? Seems to be the recommended method for paste on that particular setup.

Anyways not something that will make a huge difference, but could help in a small way.

Also 70C isn't a bad max temp tbh. Also if you are running stock settings I would say your Vcore is still pretty high. At stock I would think you could run that CPU at only 1.185 or even lower, but 1.185 would be a good starting point :-D.

Anyways if you do decide to give that setting a try make sure to report back on the results. Helps me narrow down settings that work for the majority.

[idata]

Your Vcore is high for default clocks.

I could use 1.15 Vcore for default clocks and if iam not wrong I had even 1.125 but to be sure its working just try 1.15 and you will see a lot lower temps

[cmack2]

thank you so much guys for your intelligent contributions - we are actually trying to solve a High CPU Temp/Temp Spikes Issue with the 7700K...

1) it is obvious the temperatures we are reporting is not with the tips of our fingers - we are using software such as HWMonitor, HWInfo64, AIDA64 and so on - they all actually give different readings on exactly the same computer processes - so i question the accuracy of the software sensor readings - seems there is a margin for error here...

2) reading between the lines - it would seem that such a highly engineered CPU as obviously the 7700K is (thousands, thousands+ man hours of development) - many mobos @ stock are OverVoltaging - particularly @ stock settings - thus leading to higher CPU Power & Temperatures...

3) i absolutely agree with so many on this forum that there ACTUALLY is a Temp Spike Sensor issue - it is inexplicable that sensor temperature data can fluctuate so wildly - even @ idle - in front of our eyes - using said monitoring software...

4) TGrable & other contributors on this forum, i would be happy to lower my 7700K voltages - with EXACT INSTRUCTIONS - i have AI Tweaker in ASUS TUF Z270 Mark 1 BIOS - i would also particularly like to explore higher RAM frequency with my Corsair Vengeance 4 x 8GB 3200 sticks...

5) also TGrable, i have Corsair H115i - that in maximum Performance Mode - provides very effective cooling...

6) i have used the latest Prime95 - however, must say that the AIDA64 Extreme Benchmarks of JULIA, MANDEL, RAY32 & RAY64 clearly provide the most rapid indications of all system CPU temperature performance...

thank you all for your time...

craig

[ICitr]

I just wanted to add 2 cents about using fan profiles to deal with the erratic fan cycling due to the temperature spiking.

I thought of that as a temporary solution and I looked into it. From what I can tell, and please correct me where I'm wrong, most software fan controllers don't have what's called an hysteresis function (where the value of a physical property lags behind changes in the effect). I looked at the fan controlling software in both the UEFI and the motherboard extra software for MSI, Gigabyte and Asus, as well as the software that comes with Corsair and NZXT coolers. As far as I can tell, only Asus' Fan Xpert 4 has a property for ramp up time and ramp down time. It's not actually a delaying effect, but I think you can achieve a similar effect.

The independent app, SpeedFan, does have an hysteresis property but it doesn't work with many motherboards most notably right now with Z270 boards. Even so, I'm not sure how smartly such a function is implemented. What happens, for example, when you're getting 3 second spike bumps every 5 or so seconds. I'm not sure that a simple lag time in effect will be so useful.

Then there is the Aquaero fan controllers and the AquaSuite software that comes with them. I think it will even wash your dishes for you over and above the gagillion ways you can organize sensors and fan profiles. But, now you're adding the price of a motherboard to your system for the privilege of dealing with an Intel 7700K problem. The Aquaero products I think are really meant for uber custom builds with many fans.

I just ordered my motherboard (Asus Maximus IX Hero) and the 7700K. One thing that has me concerned is the very many Asus Z270 boards represented in this thread. I hope I didn't make a mistake on both items.