Aw17R4 7820HK/GTX1080 : kryonaut repaste & arctic pad change

Hello everyone,

I advanced enough in the process of re-pasting and re-pading my 17R4 that I can start my own thread in order to ask questions and keep everything in one spot.
I went in the process of changing all the thermal pads with Arctic thermal pads and repasted with Grizzly Kryonaut + added the PCH mod.
For this, I used @iunlock and @Pete Light guides.

I chose to follow @iunlock for the heights of the pads :



I added a green circle to show that on my machine, this pad was missing.

Here are a few cloesup after cleaning the heatsink, cpu, gpu and application of the new pads :







And a few closeup once assembled :

 

Cont'd :







For your information, I ordered the same heatsinks as @Pete Light in order to do the PCH mod, which I then filed down to 3.9mm and applied Kryonaut + glue to hold it in place on the 33+ scotch tape.

All this for almost nothing!

I still have some core differentials : 0 and 2 being hotter than 1 and 3 and temps going above 90° on 0 and 2.

So, I tried something else : @Pete Light suggested that my pads over the VRMs at the bottom of the GPU could be too big, and because I still had core differentials with temperatures going up to 95° on max temps, I removed some height on a few pads : actually I applied the same heights as in @iunlock guide for the 6820HK.
I removed 0.5mm from each pads that were 1.5mm high, appart from the bottom far right CPU one where I left the rectangular 1.5mm one.

Guess what: it's even worse! core differentials are bout the same (around 6° degrees on average but sky high temperatures : 91 to 99° on each cores).

Frankly I don't get it:

- Assuming that the core are laid out as answered in my other thread while looking at the mobo with the fans being north :
[0] [2]
[1] [3]

Removing 0.5mm from 1.5mm pad as shown below should have closed even more the gap between the CPU and the heatsink right?

Isn't it a fact that the less height you have on pads, the closer the gap will be?



So for now, the PCH mod is working as expected, but the whole repad/repaste didn't do the trick.

What I am trying to achieve here, is to get even core temps, and not having temps go higher than 85° without LM. After this, once this is ok, I will probably apply LM but there is no point without getting even core temperatures first.

For testing, I didn't use OCCT because it doesn't put enough work on the CPU.
I used Assassin's creed Origins which is quite demanding, and ran tests at 40x with PL1 and PL2 set to 0, no Undervolt.

Did I miss something? Anyone has ideas?
I would be really grateful if you guys got any more ideas.
Do I need somehow to bend the whole heatsink unit?

 

there is actually a thread dedicated entirely to this issue. you can find it here http://forum.notebookreview.com/thr...u-core-temperature-differential-issue.805062/

i can see that you have a Sunon heatsink, though im not sure what revision that is without a clearer picture. All i can tell you is, getting Sunon heatsink and saying Sunon is better than CCI is a myth. It all depends on the shape of the heatsink which is not manufactured according to tight tolerances. normally this can be mitigated by using 4 strong tension arms but because alienware skimped on one arm, thus the infamous tripod rubbish design, caused the upper side of the CPU to have far less contact than the lower part.

There are 2 ways around it. First, you have to work on the HEATSINK metal, reshape it so that it can create maximum equal pressure on the cpu surface. easier said than done. this task is an ART. extremely difficult to get it right, very risky and you wont get consistent result everytime u remove and reattach the heatsink. no two repaste will produce similar result, guaranteed. again, due to the really stupid tripod tension arm design. i wonder who signed off on this design in alienware. that guy should be fired. dont bother asking people here on how to work on the HEATSINK (clariying it specially for @iunlock )metal. the best answer you will get is a vague description which basically dont tell you anything much. it is unfortunate this knowledge is a closely guarded secret by people who offer repasting services.

second way, use LM, if you dare to take the risk. LM is the only thermal paste i've seen to date with a high success rate of getting the core temperature differential reduced and under control. LM is at least 7-8X far more heat conductive than the very best traditional non conductive paste. that alone is enough to ensure heat gets transfered quickly to the heatsink even with a slight gap. every single success stories you find here are 99.9% LM, with almost zero success rate with traditional paste.

i have gone through more than 10 heatsinks of various revision, various stage of manufacturing, part number, disassembled and work on the heatsink countless number of times. this is from my very own experience. it is really unfortunate and unless you are willing to risk using LM, you are just gonna have to live with it. alternatively, there are always people here who take advantage of this mishap to make money from repasting services.

that being said, from your temps, i believe there is still room for improvement. cpu core temp differential will always be there, expect 10C or more, but most importantly get the hottest core temp under control so your cpu will not overheat.

 

Thanks Alex, but I do well know the story about the tripod design, bad tension, etc.

I will not use LM, not until I fixed those damn core differentials. To me, there is no point in doing this if at first, there is too much temperature differences.

Btw, my heatsink is a SUNON S-J-1, so first revision, and my machine is from April.

Come on guys! Help me out

 

People who join forum/community like this should be here to share/convey/give away knowledge or to gain knowledge... Not to pick up customers here for a service mission (the paid re-paste services). Why couldn’t same forum members instead create a how to do video or a very good step by step guide how to bend/fix a flawed TRIPOP heatsink? I expect they have done the repairs several times. A guide shouldn’t be too difficult to make as they fix more than one heatsink. Share knowledge or help each other is one of the main reason we frequent tech forums.

@speedsweep have you measured the fits between HS and die? Use a cheap soft creamy paste for the task. Put up pictures. Of both the HS and die with tested paste and how compressed pads is. Use the cheapest thermal paste you have.

 

@Papusan, no I haven't done that. I will need to get some cheap paste then because I only have Kryonaut at my disposal

How much should I apply? Like a pea in the center of the CPU and GPU then screw/unscrew the heatsink and check?

 

Don’t use a pea/blob for cpu. Rather use two very small thin lines 3/4 long vs die length. A small cross or blob for the more quadratic graphics die. Important not use too much paste during testing (measure how much you used). Remember you shall see how it spreads out and find out how HS fits. Next time it’s easier to know how much you will need for the final application.

 

i got a replacement Sunon S-J-1 back in May and i must say it is one of the WORST heatsink ever. the core temps were 20C apart and the hottest core can easily hit over 95C. non of the CCI heatsinks i got was that bad. which was why i said getting SUNON is better than CCI is a myth.

 

Nice try @Papusan . What type of brownie points are you trying to earn here and from who? For what? You should be much more mature than that and the things that you've stated are absolutely false and not correct.

Why are you making such accusations when they are not true? Service mission you say? Oh really now? I dare you to show me one post where I have ever advertised to the imaginary extent to support your false accusations...

Now that you've chosen to open the can of worms, let's talk shall we?

I've made available a guide to help people get on their way with repasting their system. What have you done?

a.) Troll the AW forums bashing bga machines etc... Making people feel bad for their purchases.

b.) Look at your sig for crying out loud? You're an adult Papusan, not a teenager. I mean come on....really?

c.) For someone who would never buy a bga machine and to be lurking in the AW (bga) forums... Hmmm... What was that comment you made above? "People who join forum/community like this should be here to share/convey/give away knowledge or to gain knowledge..."

Hmm... okay so I post up a free guide to help people, share, convey, giveaway knowledge etc... yet you're trying to discredit that with a false accusation? A bit hypocritical don't you think?

It's clear that you spend more time in the bga sections of the forums trolling more than anything. - To prove that, every time you post something you're advertising your sig with your hatred for bga etc...

.

To add insult to your injury, I have spent numerous hours helping people over skype with their repasting on my own time, so just to correct you again and to show that you're wrong, I have a whole lot of good folks that could vouch for my helping hand... All the while...

Can anyone who owns these systems that you call, "bga turdbook trash," vouch for you? - Hmmm... (Shall I remind you again... Look at your sig man... SMH)

If it's so easy why don't you make a guide? Oh wait, you'd have to buy these "bga turdbooks," first in order to even do that so the bottom line is this...

Be careful of the words that you speak and most of all, be very careful of making any false accusations, especially when they are not true, because they will back fire on you and expose the emptiness behind your walls rather than getting someone to fall for a lie.

 

@alexnvidia, to answer your question, information about working with LM is not a closely guarded secret. You can find videos on YT, read the instructions that come with the LM, or if you've missed it, look at the pictures on my repasting guide to get an idea. I'm curious to know what more do you expect? Every stroke of the brush when spreading the LM?

I think you're overthinking this... it's not rocket science to spread the LM with the included Q-Tip or a brush.

To be fair and I would think that you'd agree, but if someone can't figure out how to spread the LM, then even thinking about LM for that person is way above their pay grade and should resort to Kyronaut, a fail proof option.

 

@iunlock , you need to READ carefully what i wrote. i said information about how to work on the METAL as in the HEATSINK METAL, NOT LM, is a closely guarded secret.

Nobody is here asking about how to apply LM. what we are interested in is how to bend the heatsink in the right way.

 

@speedsweep, congrats on getting that. What I would recommend is getting some pressure paper to test the contact between the die(s) and the heat plate. If it's way off, try adjusting the arms, but be very careful not to bend too much or it will cause metal fatigue. (Disclaimer: Proceed at your own risk)

If the heat sink is way off then you'll be better off getting a new heat sink. The next best option would be to just use Kyronaut as the paste will compensate for the intolerances.

With the use of Kyronaut, you can be more generous with applying it whereas with LM applying too much can cause pump out, which you don't want.

Also, do a dry fit of the heat sink without the thermal pads to measure (usually can eye ball it) the gaps as some areas may need thicker pads / or / thinner pads than what is listed in the picture on my guide. (Not all heat sinks are created equal.)

Out of curiosity, you're removing the entire motherboard and removing the entire heat sink assembly yes?

 

To be fair, METAL is closer to Liquid Metal than it is to Heat Sink so I had thought you were making reference to Liquid METAL.

Anyhow, the bending of the arm is pretty straight forward so I don't know what secret you're referring to?

It is recommended that pressure paper is used, but not everyone can get their hands on that....

 

Re-read @alexnvidia s post. As well read what I talked about make a step by step guide. Or a video.

We talk about how to bend/fix a warped heatsink. Lm can’t fix this.

 

Thank you guys.

I didn't want to bring up problems about this...

I'll try all your advice.

@iunlock : When you say bending the arms, do you mean the 4 arms of the gpu part and the one arm of the cpu part of the heatsink? I already bent the cpu arm quite a bit (as you suggested and as shown on @Pete Light 's thread.
I think the pads I applied following your guide for the 7820HK seemed ok, but there is definitely something else that doesn't make real nice contact.

And yes @iunlock I removed the whole mobo and then the whole heatsink.
After that, I cleaned the dies and the heatsink copper plates for the cpu and gpu, changed all the pads, screwed it back with the new pads to check everything was making contact, unscrewed and added the kryonaut.
From what I could see, everything was making contact, apart from here :



It can't be seen from where I took the photo, but what I assume to be the choke that is underneath is not making contact (it looks like its the one where the 0.1mm stock pad is).

I am thinking of asking a new heatsink to Dell, in order to start fresh. Is there one that really is a better fit? From what I have read, CCI or sunon S-J-2 seem to be better?
If you have a good model, can anyone share the part number?

I'll see if I can find pressure paper, if not I'll buy some cheap paste to try the dry fit.

Cheers!

 

Yes the Sunon heat sinks are better. The gap that you see is not very important and that can even be left like that without any issues.

Perhaps a new heat sink may help. Also, when cleaning the die(s) it is very important that they are squeaky clean. Even the slightest bit of residue from the old paste can make that area of the die a dead zone, which could cause a lot of problems.

I wouldn't bend anything other than adjusting the cpu arm, but do so carefully. Doing a few dry runs with pressure paper would really help you see the culprit.

With metal fatigue there is no going back... it's like putting too much salt on your food. One way street. The heat pipes are very fragile so unless you have all the proper tools etc... I would not recommend taking chances.

 

Hey buddy,

Some more free advice for you (non-chargeable of course and open to all). Why don't you try using a squishy arctic 0.5mm pad? Check this I recently did for a client (This is what you're looking to achieve):


I used a powerful heatgun and bent it to what felt correct, tried it, assessed results, bent it again*. Note that there's no specific guidance here that works perfectly for all, as @iunlock has said, each heatsink is slightly different and requires careful amount of time and patience (and skill?) to get it right. These jobs can take a lot of time for that reason...

* Repeat until you get it right

DISCLAIMER: Do not break your heatsink bending it too much! unless you know what you're doing!

But of course I'm not able to reveal how this is done because its actually a mystical secret which involves lots of voodoo and lots of preying

 

Hear Hear I couldn't agree more with everything you have just said! Can I just say how grateful I and the rest of the community are for all your excellent work and the help you give individuals on Discord (includes me), without you, there would be no fix not only for Alienware but also for other manufacturers too.

The number of posts for your threads you create on this forum speak for themselves. It's a shame we have certain opinionated people on here ruining it for the majority. To me that's as unacceptable as people bashing unclewebb when TS 8.48 expired end of Oct. Completely unacceptable given all the time and effort (and love?) you've put into this community and it's problems.... SMH

 

Thanks for your reply in all of this and as a fellow member of this community that has validated all my points should make some people very ashamed of themselves...

The sad truth is that I bet it will go over the heads of the lost, confused and false accusers...

 

To add, when using a thermal pad to test it, you could also take a razer and carefully lift off the thermal pad and place a flash light under it in a dark room to check for thin spots/areas.

Just be very careful when removing the thermal pad off of it.

Most of the time the tolerance will be too small to be able to tell clearly from a thermal pad, which is where pressure paper comes into play as the best option.

If there is a noticeable difference, enough to tell from the results of a thermal pad, then that heat sink is likely far gone .... I'd get a replacement if possible.

 

Thanks a lot for all this!!!

I will try this Saturday (I won't be able to get pressure paper, I had a check and I don't know where I could find this : I live in France).

By the way, here is my current state regarding temperature after 2h20min gaming on Assassin's creed origins, set at 36x, PL1 and PL2 at 0 in the BIOS (v1.20) and 100mV undervolt in XTU.

I don't get the throttling on cores #0 and #3 especially #3...



And here is my try from the day before directly after repositioning the pads : Assassin's creed origins at 40x, PL1 and PL2 at 0 set in the BIOS, no undervolt (doesn't look good at all to me...) :

 

@iunlock snap back @Papusan lol.... appreciate all your work here on the AW forum.
Sorry i gotta side with iUnlock here that bga hate is a bit much for any grown man... a little objectivity goes a long way.
I mean i just sold my AW to get a Coffee Lake Clevo, but I also realize each has its own Pro's and Con's.
Each customer is unique in their own requirements, thus you can't just dismiss the technology as a whole.

Anyways back on topic. I just spent 3 days with 2 different mobo's and heatsinks getting my temps to line up on a 6820HK.

What I would give for some pressure paper! I had to do it by trial and error. I had the mobo replaced so I lost my last successful LM repaste. The struggle was real, I built a foam barrier and gave it a good amount of LM on the first run. No dice, core diffs and temp max floating around yours. I then salvaged a bit of the LM and repasted using 2 different types of conventional paste and reseating the HS countless times while fooling around with the pads and CPU arm. What a waste of time, nothing worked.

I finally got it on my last attempt with LM again, this time nice thin consistent layer with an extra bead in the middle. I remembered one of iUnlocks posts saying that the only way to fix these issues was using LM. Its true, I wouldn't try anything else personally. That's just my opinion coming from experience of having to tear everything apart five times. I also ended up removing the white pads and using dabs of thermal paste.

What a mess, I wish you all the best trying to drop temps with anything other then LM. Might have some luck with Kryonaut or IC Diamond.
I posted the laptop for sale while doing the repaste, sold it the second I was finished. Bye Dellienware.

That 120hz Ultrawide tho

 

Yup, good points

Sent from my SM-G935F using Tapatalk

 

Core differential looks good (well a lot better) to me. You should be happy with that. Temps are a little high though so I've installed AO tonight for you for a comparison over the weekend of what you can expect on LM if done right

Sent from my SM-G935F using Tapatalk

 

What your saying is that those are the best temps he will get using conventional paste. Interesting.

 

Those temps are hardly any good. what pete meant was the core differential is not too far apart that's all. it shows barely any good contact for the entire heatsink surface on the cpu

 

EDIT: I mixed up his before and after pics.

 

I think i offended quite a number of people with my direct comments. allow me to clarify.

true, having done all this by myself i know first hand how difficult and time consuming all this can be. those who are doing it deserve to be paid. i have no qualms about that.

Don't get me wrong. I RESPECT people who offer repasting services BUT still willing to share DETAILED AND UNRESERVED TRANSPARENT info on how and where to bend/work on the heatsink. but as we can all see, they are far too few in between. guess giving partial & vague info is better for business. it is not a sign of generosity but more of a lure to get people involved in repasting only to find out how difficult it is to get it right on your own and it's better off just sending it to the people offering repasting services in the first place. there are many in this forum who fell into this tactic.

people here need to open their eyes and look past the fancy guides which honestly dont tell you much (you can get the same with alienware service manual) and if you study it closely is not even accurate. following those guides blindly will get you into more trouble than solving your initial problems. i learned this the hard way which later prompt me to start the GPU stuttering thread to help people who are in the same shoe as i am. I was the first to stress that not all heatsinks are created equal and thermal pads need to be adapted to your individual system which some of the more fancy guides failed to mention (at first).

i can understand there are people who dont want to get their hands dirty or technically incapable of respasting and very much prefer to just get someone else to do it. but there are also people who wish to do it themselves or those like me who live in countries with limited resources and no repasting services available. im sure a lot of people who opt for repasting services got their money's worth and the services were good, but this is not about that. Please dont get mixed up with actual respasting service quality and sharing of transparent critical info.

 

I think it is more so your false frame of mind in thinking that there is some conspiracy when there is not.

There is no limiting info and all the info is freely given. Are you seriously trying to say that my repaste guide hasn't helped hundreds of people, if not thousands?

I don't get what you mean by partial and vague info is better for business. You're heavily mistaken here and it would be wise to snap out of this conspiracy mindset that you're making yourself believe.

I absolutely disagree with your thinking and opinion and so would others who read what you've written.

I'll give you the benefit of the doubt with english perhaps not being your first language (it is not mine either), but the way you state things is very straight forward so I am leaning toward not even giving you that benefit of the doubt at all.

And yes you do speak in a way to offend people. Don't expect to not get replies to what you write... especially when they are not true.

 

wierd that i wasn't directly naming anyone specific but you replied. my comments probably resonates within you more so than others, and no, there is nothing wrong with my english. it was meant to offend exactly the right people who fit my previous description.

 

@iunlock Can you please give me an amazon/ebay link for that pressure paper you use? and will you add me on discord (wolf#9584)? I need to talk to you. Thanks.

 

Really?

Maybe I wasn't clear but both Screenshots are after the same repast!
To me, having those max temps going up to 99°, average with 6° difference and 3 cores throttling at around 700mhz with a 40x over lock doesn't seem good at all.
Ok, the screenshot at 36x doesn't look that bad but still, there is throttling.
Doesn't it come from a badly aligned heatsink?
When I look at @iunlock post showing his repast with kryonaut, he has far better temps.
I mean, kryonaut is a good paste. Shouldn't I expect temps to be max 90° at 40x? I fully understand I won't get like 75/80 buy the max shouldn't be that high right?

I am looking forward to your testing on ACO, BTW I stopped playing assassin's creed franchise after the second one because the other ones were bad, but I hope you will enjoy this one as I am, it's an effing great game!

Btw @Pete Light on the screenshot you posted showing the 0.5mm pad test for the CPU, I noticed you changed the 0.1mm factory pad by some arctic ones. Did you notice some impact? @iunlock , @Pete Light : what's under this 0.1mm withe pad? Because it doesn't make contact at all, and @iunlock you said that I shouldn't bother?

 

@alexnvidia No one has insults anyone or names anyone here, from what I can see. But people know much more than what others know. I'm sitting on a lot of info. The same does probably others (I will not post names). When (whom this applies) and where this info will be posted... We'll see.

Ask for a new heatsink. Tested what I said in previous posts?

Edit. Have you asked or found out what temp you should have with stock Dell paste? And with what clocks. If temp is worse than with Dell's paste, even with several re-paste... Change to a new heatsink is probably the only solution.

 

@Papusan , nope haven't tried anything yet. I will do it tomorrow as I will have more time (if my girls leave me be ).

I'll try to take photos and document everything during the process.

Regarding the temps, they are a lot better than with stock paste (using thermal Grizzly Kryonaut every time) and as I have read numerous time, Dell is saying that under 100° there is no problem. @leeloyd found out in one thread that the actual cap in the i7's is something like 119° so I am not counting on Dell/Intel to say : "oh yeah those temps are too high!".

So my plan is tomorrow, on Saturday to:

1- Disassemble
2- Keep the motherboard and heat-sink attached and check every pads/gaps before removing the heat-sink
3- Clean the CPU/GPU/heat-sink
4- Add remove missing pads
5- Screw the heat-sink back again and recheck for gaps
6- When no more gaps, use some 0.5mm arctic pad as suggested by @Pete Light and screw the heat-sink back on, unscrew the heats-king and check for the imprint and also try @iunlock idea of holding checking for holes with a flashlight
7- When everything seems ok, apply Kryonaut, assemble and test!!!

Once I have correct temps, and around the same core differential I already have but running at 40x on heavy load, I will think about applying LM.

Again, when I look at @iunlock repad/repast/HS mod with using Kryonaut, I can't figure out why I couldn't achieve the same temps by using almost the exact setup...

I mean, I am not looking to get that good of a result (his temps are in the upper 60s at max, but there I suspect Firestrike to be not nearly as demanding as a game like ACO) but if I could be in the upper 80's at max while running at 40x on each core, with no throttling and <= 5° core differential, I would be more than happy (I repeat, without the use of LM).

 

Beacause of the QC. Big differences between the heatsink. Some worse, some better.
Fyi. Try another order to screw down the screws if you still struggle with temp (See if other have better success with an another order).

 

Quick update before tomorrow's disassembly.

Here are temps after 2h20min ACO at 40x with a 100mV undervolt :

 

Wow, so many questions! Let me do my best here

Regarding your comparison of temps on kyronaut to iunlock, don't bother that's like comparing your football skills to Messi! You won't get the same differential or temps as he will, he is super maticulous and spends many hours just balancing the heatsink to get it perfect, and he's done this hundreds of times with hundreds of different laptops... My point was more on your core differential, its completely normal to have a 2-4C variance now and then so 6 is not bad, a good effort in fact considering the fact that the core temps are up in the 90s where the variances will be worse. The real issue here (and alexnvidia nailed it above) is that your heatsink seems too far from your CPU die, yet pretty well balanced, so all your core temps are similar (within 6C) but higher than a perfect result which is probably around 10-15C less at that clock speed. I've already told you why I think that is in a previous post.

Looking forward to ACO also, just can't put Destiny 2 down at the moment haha!

Regarding the the 0.1mm pads please remember iunlocks comment that no 2 heatsinks are exactly the same and all are slightly different. Some may think that's an obvious statement to make but not all people think this and just blindly follow his guide without checking for gaps. (I actually did this myself and learnt the hard way first time on an R4 also)
Therefore, if you have a gap there and it's not because other pads are raising it artificially, fill it!


Sent from my SM-G935F using Tapatalk

 

Sounds like a plan. Just note that you mustn't put any pressure on the heatsink itself when screwing it back together when using the pad to test for even pressure. This will give you an artificial reading.

I've used the flashlight test also, works really well

Sent from my SM-G935F using Tapatalk

 

To add, even the slightest gap can translate into big core differentials, therefore it is recommended to be a little more generous with the Kyronaut.

I've posted a whole ago about actually spreading the Kyronaut to a nice even layer and then add a little mountain in the middle of the die(s) so when the heat sink goes on the pressure will spread it evenly where it needs to balance out. This works very well.

Most often times than not, the whole bending of the heat sink is not needed and will do more harm than good, most of the time due to metal fatigue. When metal heats up it changes form, therefore, even the slightly adjustment can translate to a lot. Hope this takes sense.

Also, having all the proper tools for basic metal work is important as well.

::iunlock::

 

@Pete Light , @iunlock , ok thank you guys for the last pointers before I start again
And yeah, you are right, I don't have the proper tools at home for metal work (ie. heating, bending stuff, etc.) so I will concentrate on having no gaps and definitely try the "mountain" trick with the Kryo

Just one last question raised by your last posts :
Assuming my core differential is not that bad, and the fact that my whole heatsink might be too high overall, can I just let's say put half of the pads heights (of course with all them making proper contact)?
Like 0.5mm where I put 1mm ones?
Won't it be a bit "dangerous" to have less thermal pad heights?

 

You're welcome. As for the pad heights, it's perfectly fine to add a thinner pad if that certain area needs it. I've had to do this on some occasions along with even having to add thicker pads in other places.

The most important thing is that there is enough pressure being applied when the heat sink is on top of the chips, because if there isn't enough pressure/force/compression/pads being slightly squished, even the slightest loss in compression can add up to a lot of loss in thermal conductivity.

If you need to make small adjustments, do so carefully. One way you could go about it is to run a bench with the bottom black cover off, then shut down the computer and while the heat sink is still hot, unscrew it and make your adjustments where you know needs to be adjusted. That way the metal is hot and it'll help in preventing metal fatigue.

The Kyronaut should help a lot...just be more generous with it and you should be fine.... looking forward to the results!

 

Theoretically yes but in practice I doubt very much it will work by just reducing every pad by like 0.5mm. it's more likely your heatsink is out of shape You have to take time and a lot of attempts over this to get this right, there's no quick solution which is why a lot of people don't bother themselves. You'll need to assess each pad individually

Sent from my SM-G935F using Tapatalk

 

OP! I hope you get your temps resolved. Nobody likes having a system that overheats, BGA or not.

I dont like BGA systems but thats just my personal preference.

 

Okay, had to do some cleaning up in here. Some posts were deleted.

If you're calling someone out, that's fine as this is, after all, a forum; but do so based on facts alone and don't resort to personal remarks or judgments.

Charles

 

Hey everyone,

I am in the testing process and will upload some results soon (btw, Assassin's creed origins is clearly more demanding than other games. I have not been able to take screenshots yet because of crashes, but in a game like Wolfenstein the New Colossus, everything maxed out, at 40x, I never reach 80 degrees whereas in Assassin's Creed it can reach 93° even at 36x)

I have a question though: I am trying to test extensively at 40x, with undervolting and it crashes systematically after some time.
I set the cores at 40 in the bios and the PL1 and PL2 at 100000.

But while running a game it crashes after some time, with the undervolt being between -120 and -150mV (it works fine at 36x across each cores with 120mV undervolt).
I noticed in XTU that I am reaching power limit throttling and I am wondering if this may be the cause? Should I increase my power limit in the bios to something more than 100000?
@iunlock , @Pete Light or anyone playing with those settings, I saw in some screenshots, can't remember which one, that your power limit is set at like 240000. Should I go for this for PL1 and PL2?

What's the relation between undervolting and setting the short and boost power limits?

 

I had mine set to 240000 but I think it doesn't matter so long as it's above the throttle limit.
I'll try and test scores out for an hour later for you

Sent from my SM-G935F using Tapatalk

 

So just did 75mins of ACO for you, results below at 4.3GHz:



A nice demanding game that like BF1! Looks like a great game also!

Must be said that I'm sat in the lounge next to our fire so ambient temps are probably more like 25C+, back raised 20mm

 

The Cpu should begin throttling with too low PL1 and PL2 limits. Not crash. Put 240.000 and get rid of the limits.

 

Ok. Strange then... Because I did set it to 240000 and XTU showed throttling (it wasn't constant but just on and off for a second repeatedly) although my max CPU and gpu power combined didn't reach this value.

I guess the reboots came from the CPU not liking the undervolt.
It's OK at 36x with - 120mV but not at 40x (where it seems it can sustain only -100mV).
Can the temperatures and the throttling + an user voltage being to high be the cause of this?

@iunlock, the Screenshot you posted with your temps using kryonaut, was it after playing 30min overwatch?

 

If you're using TS I'd Uninstall Xtu all together making sure to rid all Xtu files off of the system using revo Uninstaller.

Only use TS or Xtu but not both.

As for the 240000 those were just numbers that I choose from the beginning for giggles as it didn't really matter as your cpu will never reach those levels, then people started just using that number lol.. You can set it to 2345678 of you want.

My Benchmark for any games is at least an hour. Anything below that I don't really take seriously. Some games however can reach their peaks faster.

::iunlock::