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

I really hope Dell this time manage create a proper 4 screw design for the cpu side of the HS for their best BGA machine. The 3 screw design is a joke. I expected much more than this *screwed, LOL* design by Alienware.
Regarding 2 ram slots... Not many of the other OEM's put in a wimpy 2 ram sticks in their most expensive 17,3 inches "high end" laptop models. Alienware has it before as you should know, same as all other OEM's high end laptop models.
What would happen with Dellienware if not the memory manufacturers have designed 16GB sticks for laptops? O'well
There is a difference between a small 13,3 inches vs. 17,3" laptop regarding how many ram sticks you should implement. The size of the laptop chassis is more than big enough for more than 2 sticks Or maybe I have wrong
Alienware went the step from 4 to 2 ram slots in their models. Maybe the next and final step is soldered ram? If I remember correctly... Razer who prefer thinner chassis design walked this way. So I'm pretty sure Mr. Azor and Alienware will go this way as well!!
When you can't put in more ram sticks, why not soldered the fully package ram on the motherboard? Make all motherboard the same, only with different soldered graphics. Alienware has done this before as you know. Oh' yeah. This feature will come!!

 

Mr. Azor's/Engineer's thought process
"A Tri-pod is just as stable as a Quadra-pod let just use three screws on the heatsink hell that's a 25% savings on fastener's...Man I'm a GENIOUS!!!"

 

I swear the timings of their 2666 ram is 19-19-19-43

 

No no no.you all are completely missing the point.it's definitely the illuminati intervention. Connect the dots, those represents the triangle, die in the middle is THE EYE!there are specials energies flowing thru triangle shaped objects, these energies are causing your overheating, uneven core temps, etc. I don't think that you can beat this with any liquid metal, lidding,lapping,pads ...anything.those lucky ones with balanced units somehow achieved the disconnection from the main energy channel.maybe they raised a lot of prayers to our Lord during assemble process, lives in protected area, really don't know.
Only option is to buy 4pod!


Sent from mTalk

 

I don't know if you can obtain 2666 ram sticks with 20-20-20-43 If so, I know what would be offered

 

Hello again everyone!

So since my first paste/pad attempt it seemed like my temps did not help... So I went with iUnlock's suggestion to bend the #1 post of the cpu up a bit and I have MUCH better results now. This was taken on full load using OCCT and I am very happy with the results now. The temps are still uneven but it's not throttling now and that was my bigger concern.



Thanks again to iUnlock and I can confirm this does help 100%.

 

You can't be happy with 25C between warmest vs. coldest core? A disgusting result. And 93C max temp with a fully locked low powered 6700hq gradually running at 29x under 100% load is awful.

 

Yes, im happy with the results now. Its not throttling. After things stabled out the difference was only 10c which I can live with. I am happy now.

 

You may be happy. But the results are not good. This will fail (fully) on you sooner than later.

 

I have to agree with @tilleroftheearth , 10c differential is actually the exact cutoff point that is required for Intel to take a CPU back on warranty as failing specifications.

We know it's not the CPU this time, it's the design of the cooling system - it's easier to assemble it with failure as the result than success.

You made progress, but you might go back and do it again - read about the shim method besides the bending method, or just send it back to Dell as a bad unit.

That's probably the best choice, get Dell to fix it or replace it.

 

I totally get that. Do you have a link to the shim method? I must have missed it, but I am willing to give that a shot. I already have had a replacement done so I am going to hang on to this one until they have a real fix for this problem and will try to do what I can instead for now.

Thanks!

 

Google copper shim youtube
I would be careful with the shim...if you combine 2layers of paste on the bottom and top of the shim plus shim height, then i would say you can create an excessive pressure on the die if you dive the screws too much.Also if the shim jumps off the position and makes connection and short circuit some part of cpu, you know what will happen.the second scenario is moreless not so frequent.
I know that guys will disagree with me, but they are repasting their machines as often as you make the cup of coffee.In such case you are fine with some better thermal compound like Thermal Grizzly or Gelid extreme.compounds are not very recommended if you do not performing the pasting exercise frequently.
For long term application it is probably better to go with some higher quality therm pad like Fujipoly > 11w...

Sent from mTalk

 

Thanks for the info. Hmm, not sure if I want to try a shim.

I have the 17w Fujipoly pads on right now. I also have the Gelid GC-3 paste on as well. I was thinking of trying to Kryronaut next time. I think I might stay away from the liquid metal stuff.

 

You got this number with Fujipoly?!?!
Kryonaut or gelid extreme ..give them a shot then.
If new machine try to return or open the repair call.be sure that you will reach the throttle threshold soon.it won't kill you but nobody wants the crippled machine.
Also liquid metals, stay away if you are BGA cpu system owner and zero experience with this type of compound.


Sent from mTalk

 

PM me if you want liquid metal applied.

 

Sadly Kryonaut pumps out after a certain time. 8-9 months in my case. So if you're used to repaste frequently, lets say every 6 months, it's a very good paste/choice!

If you rather want it fire&forget, use IC Diamond instead.

 

It's appreciated that you are offering the metal liquid assistance to him Kudos!
Anyhow, may i ask?
How do you clean CLU off the Die and HS?
Is there any Die surface/CLU reaction reported overtime?
Also, the isolation duct tape use to be applied around the pasted area.i presume that you r keeping the tape in place all the time with the presence of CLU in the system.What is longetivity of such tape isolation? How the tape will look after 1 year when it faces all those temp ranges from 40 to 90 degrees?
Is there same particular duct tape recommended?

Thanks for the replies, in the name of community.

Sent from mTalk

 

Special care is taken care so that the longevity of the thermal compound is maximized. The tape used is 3M brand 33+.

PM me and we can discuss further.

 

Damn. Just tested my AW15. At idle all cores are close to each other (+-3C), when I run OCCT, Core 0 and Core 2 spiked to between 75C and 88C (Changed the limits to 90C) while Core 1 and Core 3 stay between 60C and 70C (Ambient room temperature must be around 20C). Test was able to run without issues for at least 25 min and didn't fail (Wanted to get some sleep) though I think it was throttle a little. Fans are pretty quiet even at high speed. Is this what you guys are seeing?

Lets see what happens after I repaste with some liquid metal in a couple of weeks. I plan to keep it and try to see what fix I can do.

 

That's a pretty big differential, and you don't know if you can bend the arm on the 3rd screw enough to compensate. I'd return it for another one. If AW gives you any static, just return it for a refund, and reorder.

I wouldn't wait beyond the return for refund period to "see if you can fix it yourself", IDK how long the return period is but I thought it was 14 days...you might check now and find out for sure how long the acceptance period is with AW.

 

Are the

Shouldn't a good application of Liquidmetal and some arm bedding fix this? If I ask them for a refund, don't they charge restocking fee?

 

Maybe, maybe not.

They'd better not for defective product...

 

The issue is that they don't consider it to be defective.

Having said that, for Australia at least Dell have a 15 day satisfaction guarantee, with which you can return it for any reason and there's no restocking fee.

 

Liquid metal is the wrong paste to use in this situation, if the pressure isn't balanced across the heatplate then that means it's not meeting the CPU and there is a gap, which isn't good for any paste but particularly bad for a thin metal paste - it will oxidize, and worse it may drip onto the motherboard and short a part.

ICD would be a good choice it's thick and can fill the gap. It may not be enough, but it's a better choice than CLU or Conductunaut.

If you return it, just say "it doesn't suit my needs and I would like to return it for a full refund". Then ask if AW will cover return shipping. That should be enough.

Bending the arm on the heatplate sounds easier than it probably is. You have a chance of damaging the plate and hurting yourself, I wouldn't risk it unless you have experience doing such things already.

 

It is defective as per Intel (vkt62's temperature differences are from 60C to 88C between cores).

 

I will look into returning it. How quickly does Dell refund the amount?


Sent from my iPhone using Tapatalk

 

How quickly/successfully can you wrestle an alligator?

(Hopefully, it will be easier than that).

 

One last question before I decide on returning it. What are the chances that the new one might also have the problem? It doesn't look like they fixed it for the new ones also?


Sent from my iPhone using Tapatalk

 

Pretty much 100% likely. They haven't changed anything. I have yet to see any 15 R3 come out of the factory with what we would consider normal temps.

 

Don't expect that DELL wants to take a loss of money on the heatsink, which has already been paid and quality secured by Dell's engineers. Dell can not complain to the manufacturer of the heatsink. They have already sent out a lot of these flawed heatsink.. This means DELL have accepted the quality.

 

You will need to pursue.. Worst comes to worst, do a chargeback if they take ages to refund the money...

 

Dell and quality in the same sentence? Does not compute...

lol...

Ah, now it makes sense (coffee does wonders).

 

Then I think I will keep this. I haven't had any issues during normal use. Ran a few games and didn't see any temps peaking and the cores were pretty close to each other, GPU was also pretty cool also and I will be repasting anyway. Only saw the temps climb only during the OCCT test. Also ordered those Shimmies @iunlock suggested. Let's see how that goes.

Also, the 1.0.6 firmware made temp reading higher for my system. Wonder what they changed.

Anyone know what that Fan Performance Mode is in the Bios? When I turned it on, it put the fans to full speed.


Sent from my iPhone using Tapatalk

 

What? Why?

There is nothing you can do to 'fix' this. The CPU is faulty or the cooling system is broken by design.

You just took delivery of a Ferrari and because it got you to work and back you think it is fine? Even if it is running on 2 cylinders less at redline?

 

That's what it does: puts the fans at 100%

I suggest to reconsider.
My AW has been "fixed" by a Dell technician (by replacing the heatsink) and the core differential has increased.

Doing a repaste might increase your temperatures as well!
E.g. have a look at this repaste job: http://forum.notebookreview.com/thr...-uneven-heatsink.797477/page-43#post-10389504

 

Returned this. Shouldn't have to deal with this for a 2500$+laptop.
For the same price i got a clevo black friday deal with a gtx1070 and i7-6820HK

 

I just got off the phone with tech support, and showed them the uneven core temperatures in HW monitor, 66/54/62/48. I had just gotten it back from the depot and they replaced the heat sink and motherboard. Their response was that the CPU was under 100C and was not overheating. I asked if the large differential in core temps was acceptable, and all he said was that the computer passed the test (5 min XTU stress) without overheating (hitting 100C) so that there was nothing wrong. He claimed to have talked to several supervisors who all said the same thing, that if it wasn't overheating it was fine. It looks like I'll have to repaste it myself, or see how much Mobius charges.

 

For 1000 less I have a 1070 and i7 6700.

 

Are they the idle temps? 66 idle is very bad...

 

It was the max temps during a 5 min xtu run

 

Strange, I haven't seen such a high difference in core temps with the max temp so low. Still, if you've got a max temp of 66C with the CPU maxed out, then I'd say it's actually quite good, and I'd be happy to keep it. I'd probably test for a bit longer and use something like OCCT or Prime95 though to be sure.

 

Try running prime95 8 threads, your high core differential should cause thermal throttling on the hot cores.

XTU benchmark and stress are ok for first testing, but not very demanding.

 

Preserving the image of the Chat with AW, in case it "disappears"

 

I'm looking at getting a new laptop and am between an Alienware 17 R4, an MSI GT62 Dominator, and a Gigabyte P57Xv6. From what I've read here, it seems that almost all Alienware 17's have the uneven core temperature issue? Has there been any more reported issues with the MSI GT62 since the first mention on the first page? So far, I don't think I've seen any mentions of Gigabyte pcs experiencing this issue?

 

And I expect you tested with Dellienware's Promised 4.1 GHz OC profile in the bios?

Edit. Pick up Wprime and Cinebench R15 benchmarks and try. I don't think your processor stop around 66/54/62/48 in those benchmarks with Dell's promised OC profiles. For Wprime 1024M stress test run as adm and select 8 threads. CPU throttling in those tests isn't acceptable. Same with to big difference in temp between cores.
http://forum.notebookreview.com/threads/phoenix-software-updates.794490/
Test also with OCCT and Aida64.

 

So, after two weeks which my GT62VR (20C temp offset between core0,2 & 1,3) has spent in the Poland MSI service center, i recieved an update from 1st line support guy, who arranged the RMA for me because it was impossible to raise RMA request via the web form.
+ 1st line is really prompt on the reactions
- no regular updates(only on demand) and web tracking codes (repair,test,quote) are saying nothing about the repair status.

This monday i asked: So what's the status?
They replied: Engineer does not know where is my problem.So i rephrased and explained again.

Today wrote 2nd time with update request in the 2weeks.
His reply is that the enginners found my unit behaviour pretty normal and they are going to it back to me tomorrow......g0d d4mn 1t!$&#&&#*#*?;
I attached 4page document with my problem, test references, screenshots,etc in printed form to the notebook.Also the same document was shared with 1st line support when we opened the RMA.

All the time i behaved extremely politely, i was not shelling them with "any update queries on daily basis", i also expressed my appreciation on their support, though after these news i couldn't stay nice.
I expressed my gratitude for this "fantastic news" they brought to me after 2 weeks...
I asked them check the heatsink contact which is obviously uneven and asked why the user needs to instruct them what to do?!?!
https://1drv.ms/f/s!AiOUCEPlhamyhftDvjpTxr9TKpGx8w

Plus i want the official statement and testing method they used.
I'm done with MSI.i refuse to argue with them...i want my unit back and need to fix it on my own.Tape,CLU maybe some shim, HS screws check...
Pity that pressure paper can be bought only in 3m length approximately for 200-300 euros.
Just sharing my exp with you.
IF YOUR UNIT SUFFERS THE TEMP OFFSET, IMMEDIATELY THROW THAT CRAP BACK TO RESELLER AND GET YOUR MONEY BACK!
This was expensive lesson for me, but positive thing is that you taught me a lot in last months.thanks to @hmscott @Papusan @Talon @iunlock @Diversion and all contributors who are not mentioned here.

UPDATE 5th Dec)
I recieved the unit today.
The repair report contains a brief information: Repasted.
That's it.
The offset is relatively smaller now, 8c in max on prime at 36x @ all cores.No throttle even withe the prime.Will check 40 tomorrow - just curious..
Will make a note on this and compare it in a few month, if grows significantly then will check the screw sockets, HS screws and will try ICD7 per your recommendation or Kryonaut combined with some copper shim.
It's a time to enjoy the machine a bit finally.
(bought in september, 1,5 of a month spent in the service centers, the rest of the time i was seeking the flaws only)
The machine arrived without any external damage.haven't checked the guts so far.
The promised statement on the temp offset has not been delivered.
CASE CLOSED...


Sent from mTalk

 

Yeah, it's a painful lesson, but if you can benefit from it in the long run, it all works out

If you can still return it that's the best idea.

If you can't and need to work on it yourself, then use a thick paste like ICD and get a "feel" for the pressure attachment placement - it will be more feel than eye-balling - though that helps too. Don't be in a rush, get to know it while you have it apart.

It may take some experimentation, you'll likely not get it on the first try, so go in being patient and look at it like a growing experience.

Good luck, and let us know how it works out

 

I maybe can return ( buyback program present) but will loose 1/3 or 1/4 of the value.it was bought in Sept.

I am afraid of ICD...i m considering Kryonaunt or CLU.
Anyhow i won't apply metal till i get proof that i fixed the heatsink.juast once i achieve 5c delta, then i will apply high quality compound.
Also wondering when the GT62VR serie component will become available on ebay, or other portal...why the heck i cannot buy standalone component?!i plan to replace the HS entirely in the worst case.

Sent from mTalk

 

If you bought in September it's of no matter, you started RMA work with them before the return date, right?

So they should offer you a no-cost refund - no restocking or discount fee.

Try to be nice and negotiate with a manager, explaining the problem that MSI couldn't figure out how to fix, and that you can't live with it without being fixed.

ICD is "scratchy" but CLU stains the die / heatplate too - all TIM can discolor so you can't really get away from it.

CLU needs a good fit, and placement on the die and heatplate at the meeting point, and is tricky to use.

It's all going to be a pain to do, and redo, and redo again when it pumps out or oxidizes in 6-12 months...

 

What circumstances are you seeing this large delta in temps between cores? 20C is definitely extreme and would immediately point to poor heatsink contact or a improper paste job (lack of paste).

In a lot of games you will definitely see temps differences because the cores are not being loaded equally. BF1 for example shows load deltas of 20-30% sometimes for me and this can lead to slight temp deltas on the cores. Again this isn't surprising as the cores aren't being abused equally.

In a synthetic benchmark where the cores are being loaded up equally the temp deltas should be much tighter in a ideal situation.

I hope you get your laptop back in the same condition you sent it off to MSI in. RMA in my opinion on laptops has never been good. Many times I have read about people getting them back dented, scratches, broken plastic areas and loose fittings. A tech doesn't give a **** about your laptop or its condition and they know they're not going to be fired if you complain. Good luck. Next time return or replace the laptop with the retailer if possible.