Warning: Some i7-6820HKs and i7-6700HQ have Uneven Core Temps due to Uneven Heatsink

@hmscott http://forum.notebookreview.com/thr...nware-17-r4-15-r3.797378/page-6#post-10373655
I don't know what to say anymore 6700Hq or 6820Hk in AW machines doesn't matter.

 

Any CPU can has potential problem, it's not limited to only the 6820HK, it can happen to any of the current CPU's, including the 6700k...

Everyone should check their CPU's. I always do. It typically has been a very low percentage of a CPU manufacturered with that fault, but enough of a number to check each time I get a new one.

I would expect everyone that see's this thread realizes it's not just the 6820HK, it can be any CPU, this thread became a point of focus when we noticed a surprising turnout of "hot" core 6820HK's posted.

Check all your new and old CPU's

 

I am currently wondering whether this fault just affects the temperature probe (which would result in a CPU swap being necessary but nonetheless...)
My reason for this train of thought being that my CPU goes from 4X°C to 9X°C in less than 5 seconds once load is applied, which seems to indicate a defective cooling system or non working thermal compound.
However, there is no "final" temperature for me at which the cores stay!

Even under a continuous load (Prime95, Dell stress test, ...) the temperature will hover between 8X°C and 9X°C, only staying at each temperature for one measuring interval.
Additionally, the different tools that read temperature (I've tried AIDA64 and HWMonitor) are consistently showing that the cores will jump wildly in temperature at a moment's notice.
E.g. one core will display a temperature of 75°C, the next interval will show a temperature of 93°C, then 82°C, ...
This seems impossible to me due to thermal constraints - I can't imagine the core heating / cooling this quickly in such quick succession under a continuous load.

To test this theory, I have run AIDA64 and monitored the 4 cores' temperatures while being idle, measuring once each second.
Have a look at the screenshot - this is definitely NOT normal.

I have contacted Dell's service and run through the usual built-in diagnostics, which didn't show any faults...
However, I was asked to provide them with my measurements / screenshots and they will contact me again on Monday to discuss further actions.
I'll keep you posted

 

Never attach any images in nbr, give links to other websites. (nbr has image cache server problems, some of they are awake only at specific times)

what is the model of your laptop ?
Are you on stock clock?
One probable cause is that AIDA is not working prorperly. Uninstall & Get AIDA Engineer.
Your CPU usage should be 100% but its pretty low, if its not from aida then this seems like a CPU fault .Maybe as Dell stated, the CPU is not giving proper diagnostic information.
Use OCCT/hwinfo/prime95/cpu-z/AIDA Engineer and post the results here. (links to image hosting websites)

 

It's an Alienware 15 R3.
Yes, I am running the system at stock clock and the CPU is idling.
So there should not be any usage, which AIDA64 also agrees with.
However, the temperature is bouncing up and down anyway, even though it should be pretty stable in this condition...

Dell has yet to state anything.
So far, they have just requested my screenshots / measurements.

 

So you are not stressing the cpu while tests??
That statement was from AW_Umar regarding the CPU problem.

 

Yes, I am stressing the CPU while testing (especially if I am interested in the temperature differential and the maximum temperature).
However, this screenshot (" AIDA64 idle.png") does show idle temperatures - so there is no stressing of any component involved (besides my nerves )

 

Yees I know. Here below more of the same shiity from DELLIENWARE http://forum.notebookreview.com/thr...te-guide-results.797373/page-25#post-10374532
This time only a pitty 15C degrees temp difference between core #0 and core #3

 

Guys I am here to talk about a few things i checked and found them to be unusual.

Unlike many people on the inter webs, my Alienware 15 R3 did not show symptoms of having any sort of electronics disease. But i was wrong. Because my laptop did not thermally shut down, I felt i did not have that problem.

At the same time, i did notice that i would get lags and hangs here and there, specifically when a transparent window would be displayed. The whole computer would hang and my mouse would not move and seemingly everything stopped for the moment.Even event viewer would not be able to express such an error. But i was plagued with the dwm.exe -desktop window manager- stopped responding.

The fix for this was completely stopping the intel HD 530 driver and device (using Fn + F7) and thankfully this option is available in our alienwares.

What i did next was to go and properly check my temperatures. I was running at 4GHz OC before this test and had played Witcher 3 and BF1 without any issues (hence my belief that nothing is wrong with my laptop), so i only expected victory. My GPU temps, because of me using MSI Afterburner, were known to me and were 69C max. But i never paid any attentiuon to my CPU. How wrong is was....

This is my CPU graph on load: http://i.imgur.com/xIroQpB.jpg

As you can see, there is a gap of about 20 degrees between the pairs of cores. Which cannot be good.

So thinking, actually fearing, that i might have to get this sexy machine replaced, I got myself some Arctic Silver 5, repasted, and performed the tests again. To my disappointment, and to the wisdom of those on this thread:

http://forum.notebookreview.com/threads/warning-some-i7-6820hks-have-uneven-core-temps.797477/

it did not work even a bit. In fact this graph is with the paste on.

But even after this, there is a particular issue that has seen very little time in the public. I think it is the general lack of vocabulary for this issue that would make the expression of the problem so difficult, it cannot be searched for, and written about properly. This would result in two people having the same issue but different words for the issues. But here comes something very very peculiar:

This is the graph of my GPU when running both the intel HD 530 and the gtx 1070.

http://i.imgur.com/YN10IWn.jpg

It is very peculiar behavior, and i encourage everybody to install MSI afterburner on their laptops and check for this. As far as i can see, and you can too, by the graph,

http://i.imgur.com/FTixCRc.jpg

It stays onliune for a second, and the next second it turns off. Who do i say off? Because no sir, the temperature doesnt jump between 0 and 49 for example. When it says 49, it is 49 but just the next moment it is a "--" which reads it to be an unreadable value by virtue of the device being shut off completely.

I have decided to upload 3 videos of the same: https://drive.google.com/file/d/0Bzj3rWonpXzaWkJOTHZEejRNUE0/view?usp=sharing

https://drive.google.com/file/d/0Bzj3rWonpXzaakZGU3VxS0ludEE/view?usp=sharing

https://drive.google.com/file/d/0Bzj3rWonpXzaN2QxSjVXY2ZsT0k/view?usp=sharing

Please not this: I faced no issues while gaming. The GPU stood at 69C Max guys. This is a hell of a laptop as far as i saw its performance in gaming and temperatures of the GPU. The only gripes i have are the hangs and stutters. It also feels like these would cause DPC latency issues but i am in no mood to check for it anymore.

I have talked to a dell Rep about the GPU stutter, but not about the CPU thing. I will be telling him about my readings and get a new alienware 15 R3.

My system: Alienware 15 R3, FHD display. i7-6820HK. I have uninstalled software like Support Assist and DFS (Dell foundation services). I have performed clean installs on my system.

Regards

 

Thanks for the awesome detailed info

I assume you have been posting this in AW NBR and other AW forums - if not please do, as that's where it will do the most good.

If you can also link back here for the "Hot Cores" Issue in your posts - where the CPU cores with equal load have a high temperature differential - some cores 10c-20c hotter than the other cores. Like you see even after repasting.

This Hot Core issue also affects the 670

AW will need to replace your motherboard, or swap a new laptop with yours - the only ways to get a new 6820HK.

Please come back and let us know how it works out

 

Came here to report 85-90C temps on Core 0 and 2 on my 6700HQ on my AW15 R3

 

That's nice, but we are more interested in the contrast between those cores and cores 1/2.

As in Cores 0/2 are 5c higher in temperature than Cores 1/3 - the differential in temperature between Cores 1/3 and Cores 0/2... make sense?

When the temperature of Cores 0/2 lead the temperature of Cores 1/3 in higher temps, say 5c, or 10c, or 20c higher, then Cores 0/2 reach the thermal throttling temperature first - throttling the whole CPU including the cooler running Cores 1 / 3.

The way to test it is to get a test that loads all 4 cores evenly, and then monitor the core temperatures like with hwinfo64, reset the readings just before running the benchmark, then the High Temp or Max Temp of each core tells you the information you need.

If cores are all like:

85c
82c
83c
82c

That's good, all the cores are within a few C in temperature

If the cores are all like:

99c
85c
98c
84c

Then the temperature differential is about 15c, which is too high, and you should return that laptop with that CPU for a new laptop with another CPU. Either by returning the laptop to the retailer - easiest and quickest - or RMA the laptop to have the motherboard with the soldered CPU replaced - or they may also swap another laptop for yours as a fix.

Check it out, and hopefully the Core differential is 5c or under, and then it's normal

 

Oops yeah sorry, after about 5 mins on prime95
Core 0: 91C
Core 1: 69C
Core 2: 92C
Core 3: 67C

 

Apparently that's normal these days. Nothing to worry about. Just game On.

 

Unless you can cross 100C there's nothing to be concerned of.
Try to cross 100C

 

Yeah, that's a bad one.

Over 20c differential between cores.

I wouldn't waste time re-pasting, usually it doesn't work. It might bring down temps evenly across all 4 cores, but the temp differential would still exist.

Assuming that's from an AW laptop, contact Dell and ask for a replacement CPU - which would require a new laptop or a new motherboard.

If you have at home replacement you could get AW to swap the motherboard or swap the laptop in home to avoid shipping back and forth.

If you can't do that try for simultaneous cross shipment - AW send you the new laptop while you send the bad one back to AW.

Good luck, and please come back and let us know how it works out

 

That's after a re-paste...

 

100C is the thermal limit of the 6700HQ.

 

BTW, this also means the paste job on your CPU is highly uneven among the cores, resulting in the temp differentials.
At the worst your cpu will die in 2 years, under heavy load.
So just get a 3 yr warranty.

On the bright side , think about dell. If your CPU doesn't die , how will they sell their 1280 gtx to you huh?

 

Darn, and I am assuming you ran a few tests, not just Prime95?

It's always good to get confirmation from more than one test... this is more for anyone else reading this, I assume you already did this

It's time to return it...

 

Oh i did, Battlefield 1 hammers the hell out of it. Same temps. It's across the board.

 

Please don't do that, that's horrible advice: TJUNCTION 100°C
http://ark.intel.com/products/88967/Intel-Core-i7-6700HQ-Processor-6M-Cache-up-to-3_50-GHz?q=6700hq

I know you are trying to help, but maybe sit on the sidelines and watch for a while, to see what's happening here.

He has a bad CPU with a high temperature core differential between cores. He already re-pasted, lowing the overall temps across all cores, but still has the high temperature differential between cores, causing pre-mature thermal throttling.

A long warranty with AW hardware is always a good idea.

Now he can use that warranty to return the laptop for a new motherboard or whole laptop exchange so he can get a "good" CPU.

 

I didn't know that he has repasted! If you have used repasted with liquid metal and still having this prob, then its a cpu fault and return that **** back to dell.
but if you have repasted with a dried thermal paste, this might happen (and also if the heatsink is not properly attached)

 

In the case the 6820HK's and 6700HQ's Intel seems to have put out some bad batches, and these CPU's have an internal failure showing high temperature differentials between cores.

Re-pasting won't do anything except lower the overall temps of all cores, it won't fix the core differential.

The first time it happens to anyone, they will try re-pasting several times - I know I did - but it won't help

That's why we made this thread, so that people could be saved from wasting time re-pasting, and get their laptop returned for another one from the retailer - instead of running out of time for a return - and possibly getting stuck with a broken laptop.

Thanks for your help

 

You are ruling out the possibility that there maybe hardly any stock thermal paste over cores 0,2

 

No, I'm not

Again, it's difficult to wrap your head around this the first time. We all have run across bad paste jobs, and re-pasting with good coverage fixes most problems.

These problems are internal to the CPU.

My first failed CPU like this was a 980x. Re-pasted with a couple of pastes a few times, and the core temperature differential stayed the same, as did temps - my first paste job was already good.

After taking the CPU back to Microcenter and getting another one - that wasn't easy, BTW, I put in the new 980x and it worked perfectly the first time.

Same paste, same Noctua cooler, everything the same - except the CPU.

It's rare, but I've had a couple, so I could recognize these recent 6820HK temperature profiles as being just that, internal failures of the CPU.

It looks like some 6700HQ's are also failing.

This is happening to other makes as well. MSI has had failures like this recently too.

I am assuming that the 6820HK failures have been noted by Intel / vendors, and that's why it disppeared from common configurations. It's hard to get one right now.

Again, thanks for the help.

 

.... mobile CPUs arnt made the same way desktop CPUs are ... the issue with the 980x you are referring to is a TIM issue between the core die and the heat spreader. Mobile CPUs dont have a heat spreader and the heat sink (if it is making proper contact across the die) interfaces with the top layer of the CPU die. any imperfections or layer separation below this layer would result in a dead CPU.

 

It wasn't the heat spreader either. I didn't want to complicate things, but delidding didn't solve it either.

The failure is internal to the CPU die.

Under the heat spreader the desktop CPU looks like the mobile CPU - which comes as a bare CPU without a head spreader.

It could be many things, including bad sensors. A die failure of some kind.

 

so wait ... you delidded a 980x and microcenter still exchanged it for you?

 

It wasn't an easy return. So I don't do that any more. When I find the failure is a core differential now, I don't waste time trying to fix it myself, I leave that up to Intel

That's also why I wanted to start this thread. This problem can cause a lot of wasted time and angst for those involved.

Use your warranty and return the laptop - don't even bother re-pasting if there is such a pronounced core temperature differential. Save yourself the time and disappointment(s).

 

I have contacted Dell via their support hotline and have run them through my results.
They requested some screenshots and promised to call me today (Monday) in order to agree on further steps.
They pointed towards a faulty cooling unit or defective thermal compound and promised to fix this on-site - as it stands, I have got an appointment with the technician on Wednesday
I am quite curious and will keep you posted.

 

We confirmed that this is an issue with the heatsink/mounting hardware...i'll let @dadealus chime in with the details...

 

Wow, that's great

You have even temps on the cores now? What is the differential?

Or are you guessing based on some theory?

I would suggest sending it in to get it fixed by AW so they know what the problem is so they can fix it in production, so this problem stops happening to so many people.

We've already had several people try to fix this with re-pasting, and although all the temps drop, they still had the core differential temperature remain essentially the same.

 

Bad news -- did you notice that the CPU block only has 3 screws? Well it leaves uneven pressure on cores 0 and 2. We tried reapplying pressure to those cores (and overall testing different types of pressure on the block) and discovered it. It looks to be a perm issue unless they can find a way to re-tension one side of the CPU.

 

So, it's not solved, it's just variable based on pressure between the screws? Not convinced this is the problem, but if you are able to change the core differential it might be related.

That's interesting, as re-pasting - and re-applying pressure to those screws could cause the problem if it wasn't happening in the first place?

I would suggest letting AW know what you found and send it in to get it fixed by AW so they can fix it in production, so this problem stops happening to so many people.

We've already had several people try to fix this with re-pasting, and although all the temps drop, they still had the core differential temperature remain essentially the same.

Not all new AW laptops have this problem - owners have improved great temps after re-pasting - which requires loosening and retightening those same 3 screws, and MSI laptops are also showing the problem, for now I still think this is a CPU die failure overall.

Is there a service manual yet for this laptop? Maybe there are notes on how to torque down those screws for proper alignment / placement to avoid this problem - if this is really the cause.

This is also happening on MSI laptops, I would be surprised if MSI has the same screw layout / alignment - and the Core 0/2 differential happens with MSI laptops in the same way.

It would be awesome if it's solvable with a new heatplate/base using 4 screws for even pressure. But wondering what the solution is with the same symptoms on MSI laptops?

For now, still treating this is a CPU die failure, but hopeful it's bad cooling plates from AW / MSI, and the solution is the same: return to vendor for a new unit or AW RMA / in home service.

 

Not solved, and won't be solved. This is after 3 repasting and testing the results. When you apply pressure on the block where the 4th screw should be, that's what evens out the temps. I find it very odd that they opted for a 3 screw design for the CPU on the R3 and a 4 screw on the R2. That's prob why the R2 had better temps on the CPU side of things.

 

So you are holding down the heatplate where the 4th screw would be, and running the test(s) to get readings??

Did you happen to get hwinfo64 core temperature A/B screen shots, one without applying pressure and one with applying pressure?

Is there a service manual yet for this laptop? Maybe there are notes on how to torque down those screws for proper alignment / placement to avoid this problem - if this is really the cause.

This Core 0/2 differential problem is also happening on MSI laptops, I would be surprised if MSI has the same screw layout / alignment - the Core 0/2 temperature differential happens with MSI laptops in the same way.

Not all new AW laptops have this problem - owners have improved great temps after re-pasting - which requires loosening and retightening those same 3 screws, and MSI laptops are also showing the problem, for now I still think this is a CPU die failure overall.

If you have the problem, reporting to AW the core differential temperatures and this 3 screws applying pressure unevenly idea, that should get you another laptop - one which might work.

Or, return it for a refund and wait for AW to fix the problem in production.

It would be awesome if it's solvable with a new heatplate/base using 4 screws for even pressure. But wondering what the solution is with the same symptoms on MSI laptops?

Thanks for the info, please come back and give us updates as AW works with you for a fix.

 

I do not know what people with only "three screws" on their cooling system mean by the results changing, when they apply pressure to the third screw receptacle. I have 4 screws in my cooling mechanism, and i secured them pretty tight, making sure that the CPU and the GPU got good contact with the heatsinks.

Guess what? Lower temps by about 5 degrees, but the 20C differential still exists

My CPU throttles, one core, while playing Witcher 3. Thermal and confirmed by HWinfo64. let me know if you wantt me to post ascreenshot of the readings.I repasted with arctic silver 5.

P.S. you can read more about my problem a few pages back.

 

Folks...we have another one....sure enough 15R3....15C+ difference cores 1 and 4.

@Mobius 1, can you post up that screenie?

::iunlock::

 

Was looking to @Vassilis008 's system when this happened

 

Ran at 3.8GHz @-80mv

 

Another one in the bag....so strange....what's up with these crop circles of temps lol....

Could it possibly be that the tension screw holes (pieces) on one side of the mobo is lower than the other side, thus causing unevenness?

Maybe it's time to test with copper shims....

 

You guys all have fun wedging and tweaking the pressures on the mountings for the heatplate / base, maybe washers between the base and the plate, or some other alignment adjustment will fix it.

For others that just want it to work without pulling apart their brand new laptop, return it to the vendor, get another one, and let them figure it out for you.

You paid for a working laptop, let them know they failed to deliver one, and to please send you another one.

Whether it's bad cooling hardware alignment or a bad CPU, a vendor solution is advisible. If as part of a self repair you bust something, that's on you; things can get worse.

Unless you want to spend a bunch of time trying to fix it on your own, which I can tell you from past experience might not end up fixing it. You might be wasting your time.

GL, and please let us know your progress, whatever route you take.

 

I agree. If the issue is really bad, which a 15C+ in difference is, then just get it exchanged and don't hassle with it. Not only is it wasting valuable time, but it's also counter productive in that the OEM's are not having to deal with the issue...the more units they get sent back due to defects, the better for our sake. Let them deal with it so that they can also improve the QC overall.

The only exception is if one has a silicon lottery winning chip with an actual thermal paste issue...I can then justify trying to make it work but.....that's it otherwise...send it off.

 

If you have a +20c core temperature differential, you won't be able to OC far enough to know you have a Golden CPU

 

True true. Hypothetically speaking.

In some cases where it is a thermal paste issue, the temp differences would show up gradually, but in this case where there is clearly a hardware related issue....no can do.

 

I also have a 9c temp difference on average on load with my AW15R2 with a 6700HQ. It was less apparent while I used IC diamond 7. But with Kryonaut, I have the problem that the paste gets pulled away from the vacuum pulled by the heat where there is more space under the heatsink. I have requested a new heatsink from Dell. Hopefully it arrives soon. My heatsink seems to be slightly warped. It doesnt sit perfectly straight on top of the CPU which causes the problem in my case.

 

I was hoping someone would post some photo's of the problem, with some annotation of the issue, and what they tried to do to fix it - and their results.

Do you think you can rig a fix yourself, or does it need new components from AW for make the repair?

If you do get the parts from AW, could you please do a step by step on the problem, and repair - with photo's, along with before / after hwinfo temperature screen shots showing the improvement in core to core temperature differentials?

Thanks!!

 

Yeah I will do my best to gather some information.

To be honest I do not think myself its the fault of the chips. There are some temp variations between the cores. But there are so many factors where it can go wrong by with the heatsinks. Its hold down with about 8 screws with springs between the screws, heatsink and PCB. All these screws need to have even pressure, the thermal pads are all around the CPU and GPU dies asymmetrically and also put some pressure on those springs by pushing them back up again away from the CPU and GPU dies. Then there is warping caused by the heatpipes that connect 2 pieces of heatsinks together etc.

Add small inconsistencies on manufacturing between all these parts together and you got a recipe for not perfect fitting heatsinks.

 

I'll be repasting two 15R3's this week so I'll for sure pay extra attention in seeing any discrepancies with the heat sink.

Just when it feels easy to write it off as a hardware issue (the CPU itself), the possibilities of all that you've mentioned brings things right back to square one.

Hmmm....what has made me think this could be heat sink related is that the issue seems to exist across the 15's....I highly doubt that they cherry pick certain 6820HK's to be used for the 17's vs the 15...

Who knows...haha...we'll get down to the source of this soon enough.