[Alienware 17R4 / 15R3] - Disassembly + Repaste Guide + Results

Sorry,
Referring to PRIME95 small FFT "stress testing" with AVX disabled
(CPUSupportsAVX=0 added in local.txt) disables AVX and FMA3.

Or AVX enabled
(CPUSupportsAVX=1 and CPUSupportsFMA3=0)

 

Does HWinfo64 have AVX on or off when it monitors the CPU cores ?

 

AVX has nothing to do with HWinfo64.
it's a CPU instruction extension that uses more heat/power.

 

Oh, it was absolutely my fault. So far, so good with my DIY effort to keep the cable in place.

 

Hmm, I just bought some LM, I really need to get this done to my 17R4, but how risky is it do this myself? I can build PC's etc fine, but I've not gone into watercooling or case modding... on a scale of 1-10 how hard is this to do for a casual PC builder...?

Thanks.

 

Applying it properly, and just as important, sealing it properly would probably be a 9 for a casual PC builder. And, the risk is considerable if you don't do it properly.

 

I did the repaste. it's all gone swimmingly!! Temps are super low and all is well, but I can't keep my overclock due to power limit throttling - this sounds like a setting in the bios (though I did install intel thermal drivers...) anyway is there a setting I can change to remove the power limit?

Thanks.

Thanks for the info, I quite agree - though I did it anyway and it's all working like a charm!

Question is, I know some people have passed the 4.3Ghz mark, what settings are being used for a 4.5Ghz for example?

Thank you.

 

Oops, don't know where I read 4.5 or something, but it seems 4.3 is the max - with that being the case, after a 2hr stress test it's fine with no throttling of any sort and reasonable temps. Very glad I did it.

 

Just make sure the core temps are within 1-2 degrees core difference. Anything more than that and it can show signs of drying out as the contact needs to be almost perfect.

 

Thanks for that info, my understanding LM needs to be reapplied periodically?

I guess I'll just keep an eye on the temps and repaste as necessary....

 

Been about 3 months, and my laptop is still alive despite taking it to work and home daily (60 mile round trip commute). Temps look better now than they did before! I even turned off throttlestop and I'm only getting about 70 C max.

Sorry I haven't posted much been too busy learning about my Model 3 .

 

Guys, i need some advice here, i'm pretty far away from home (in the middle of the sahara desert for an Oil&Gas Project) but i have with me my little survival kit (conductonaut, MX-4, cotton buds, isopropil alcool and surface purifier, 33+ tape, set of precision screw drivers, plastic prying tools... well basically what everybody carry with them, isn't it? )

So, when i first got the 17R4 i've applied Conductonaut on both GPU & CPU. Temepratures were really good. GPU maxed out at 55 degrees and cpu on combined load went up to max 65 (7820Hk at 4.2Ghz) with a core differential of around 1 - 3 degrees.

Last week temperatures on Core 3 started to rise and 2 days ago, when it started to show over 10 degrees difference from the other cores i decided to openit up. Basically what i found is that a portion of the CPU was completely without any trace of LM and instead it was "embended" into the heatsink. I cleaned up and repasted, this time with MX-4 for sake of seeing what difference it makes.

Now, while on the CPU my temperatures now are way higher (75 / 77 degrees) on the GPU side i got no difference at all (!). Yeah, my GPU even with MX4 goes max to 55 - 56 degrees while gaming for some hours.

I was wondering, what if i repaste the GPU keeping MX-4 and LM just the CPU? i'm just worried that if improper ammount of MX4 is applied to the GPU then the CPU might suffer from poor heatsink contact. Or i'm just being paranoid?

Thanks for the advices.

Oh... and i hate the CPU tripod mount system! who the heck come out with this idea?!

 

Haha I also always carry those things with me when im away for a longer period of time.

The thing is LM is useless when the contact pressure isnt perfect, it slowly dries out and oxidizes, in some cases you have to scrub it off from the CPU die with sandpaper even.

In that case the safest bet is to use ICD7, the longest lasting paste and or use better thermal pads that can compress beter so you have less back pressure on the heatsink mounting to prevent improper pressure.

The tripod isnt really the issue, the too hard stock thermal pads are unfortunately. If you test run it without those thermal pads you noticed the pressure is perfect. So if you use something softer like gelid pads you can prevent the thermal pad pressure from ruining the CPU die pressure.

 

Thanks for the feedback Rinneh,

I do believe that in my case was a bad application on CPU side, the GPU was perfect as day one, on the CPU i did not really apply the proper ammount on the heatsink (thing that i did on the GPU) i just used some leftover from the cotton bud and called it good.

 

Yeah but the GPU thermal pads have a big influence on the CPU side. So thats why a repad is still recommended.

 

Hi All,

I repasted my Alienware today thanks to help from this thread. I went from getting constant thermal and power throttling in the XTU benchmark with my 6 core turbo speed flicking between 3.7 and 4.3 ghz to a constant 4.3ghz 6 core turbo speed with no throttling. My cinebench score went from around 1100 to 1315. I did this with just Arctic silver Ceramique 2 as that is what I had in my drawer. I know a lot of you have used liquid metal and bought better pads etc, but to be honest my system is now performing great so I am happy. I took a few pictures along the way in case anyone is interested. From what I can see it is pretty much identical to the R3. And yes, it has horrible thermal paste from the factory still

Pictures: https://imgur.com/a/0sZzcyu

It is probably still a little too warm to run the 5ghz overclock profile Alienware includes, but I can't really complain at 4.8ghz turbo speeds on 1/2 cores in a laptop!

 

Hi everyone, I'm a 1 month owner of an Alienware 15 R3 with 1070 GPU Micron and 7820HK (Not OC). I've read all the pages on the different topics concerning the heat problems: core differential, nvme ssd, pch mod, etc (Thanks so much to all the contributors, especially iunlock).

After reading all of this, I did the following modifications:

- ThrottleStop (SpeedShift activated, -120mV undervolt)
- PCH mod (3M super 33+, alu heatsink, Kryonaut)
- Nvme SSD mod
- 2 attempts of repasting (Kryonaut for both CPU&GPU (spread method with a small drop at the center), Arctic MX-4 GPU (Pea method & Kryonaut CPU (spread method)) - Correctly screw 360 turn 1 by 1, order 1 to 7

I didn't do:

- Repadding (should receive soon Gelid thermal pads) - I only have a ~6 degrees core differential
- Bending the tripod
- Using Conductonaut (too afraid)
- Calling Dell for having support (Don't believe that it will change something...)


I'm still facing major heating issues (90°C GPU => throttle & 90°C CPU)... BUT, I'm facing heating issues ONLY WHILE AC PLUGGED (genuine laptop charger). In battery mode, I'm getting same results in unigine heaven or in game with around -20/25°C less.

I tried everything (all the settings in nvidia control panel, nvidia inspector, undervolting gpu with msi afterburner, not using ThrottleStop, disabling/enabling speedshift/speedstep, latest drivers OR drivers from dell support website with my laptop serial number, switching W10 power plans, etc).

After making a ghost of my W10 (which is optimized as hell), I'll try to install W7 to see if the problem is still happening (which is my last option).

I'm asking you: After all those attempts and modifications, do you have an idea of what is my problem? Is there someone in Europe proposing the same service as Mobius 1 or iunlock do in the US?

1 month having this laptop and still haven't used it because I'm too afraid of these high temps.

 

Bending the arm doesnt do anything from my experience. its so "springy" that it just moves back to the original position after a while. The reason you have higher temps while the AC adapter is plugged in is because on battery the laptop cannot go all out because the lack of power from the battery. The battery only can supply about 100watts of power at a time. The AC adaptor ofcourse 240. 90c on the GPU is a big sign that your heatsink isnt making proper contact with the GPU. With these systems the GPU tends to run a lot cooler than the GPU. Even with stock paste it tends to be around 70~75c under load. So something is really off.

Also dont use Kryonaut, it dries out too quickly. Try something like ICD7.

 

Thanks rinneh for your support, I'll keep in mind the ICD7, so if my problem is a bad heatsink contact, I'll wait for the Gelid thermal pads before calling dell support for having a new heatsink. Or what do you advise me to do?

Also the thing I don't get is that I don't have a consequent core differential, so I thought my heatsink was making good contact on my CPU...

 

At 90c it is possible some cores are already throttling so a core differential is maybe not presenting.

What are the ambient temperatures by the way? Does raising the back of the laptop help a lot?

 

If I only stress the CPU with OCCT or Aida64, I get max 80°C on the CPU with the ~6°C core difference (74°C). It is when I run unigine heaven or any resource demanding game that I get the 90°C on GPU & CPU.

My ambient temp is 17°C, haven't put the heater yet... Raising the laptop doesn't make any difference at all, I even tried handling it in the air while benchmarking but same...

 

Hmm very strange behaviour. Your cpu is okay when it comes to temps but 17c is quite low for ambient temps. This is problematic when it is summer. But hard to analyse slyze the exact issue when i am not handling the machine myself. How is the paste spreading on the gpu block from the heatsink pressure. You sure its making good contact?

 

Precisely, I'm not sure that the GPU is making good contact. Here is how was the stock toothpaste: https://imgur.com/a/Xx32Ob6

I tried first spreading a very thin layer of Kryonaut and then I thought that the GPU wasn't even touching the heatsink so I put a huge amount of Arctic MX-4 (Pea method) to be sure to have a contact but same problem...

 

Hmmm very hard to say whats up. the indentations on the thermal pads from the memory chips etc arent that deep it seems......I would ask for a new heatsink tbh.

 

Thanks rinneh, before asking for a new hs, don't you think I should wait to receive the Gelid pads and replace the stock ones?

 

Well I would have expected a deeper inprint on the stock thermal pads where the memory chips are located. Its hard to check based on the photos because 2 pads are stuck on the GPU memory chips and not the heatsink. So cant see the imprint. But the one that I can see doesnt have a large imprint suggesting that the pads arent too thick and thus the heatsink is pressing harder against these.

 

Update of my situtation, I finally decided to ask for a new HS before trying Gelid pads, the one I had was the AT1JM00 5ZC0 and I received the AT1JM00 7ZC0. I didn't bother with their stock toothpaste and replaced it directly by spread Kryonaut.

After doing some benchmarks with unigine heaven (laptop flat on my desk), I'm getting (undervolt -120mV):

------------------- GPU -----------------------
- max 69°C (average 68°C)

------------------- CPU -----------------------
- Core 0 max 85°C (average 72°C)
- Core 1 max 84°C (average 69°C)
- Core 2 max 79°C (average 65°C)
- Core 3 max 80°C (average 65°C)

Huge improvements for my temps, but I have now a 7°C core differential (comparing Falkentyne cpu photo on page 335, I'm sure this new HS is not well balanced as Core 0&1 on the left side are hotter than Core 2&3 on the right side) but it's at least playable while plugged.

When I see iunlock temps on 1st page with Kryonaut while playing Overwatch, I'm still thinking I have a problem. My GPU is a GTX 1070, he tested a GTX 1080, his stock temps (62°C) were lower than mine with Kryonaut. I'm even thinking asking for a new MB

Should I give it a try with Gelid pads?

 

He used liquid metalpaste not kryonaut..... Your temps are absolutely fine. The Core 0 and ! higher temps is something I also dealt with. This is an indication that the thernal pads on the GPU side are slightly too thick and or that the heatsink is not perfectly straight.

But too be honest if these are your temperatures I would keep it without a second thought. Only Kryonaut will dry out quickly replace the paste with ICD7 then at that time.

 

No check his 2nd post on 1st page: " Grizzly Conductonaut (Liquid Metal 'LM") vs Grizzly Kyronaut (Traditional Paste) ", it is Kryonaut that he is using and his temps are awesome. That's why I keep wondering if my temps are correct. Something I haven't consider is I'm having a 15R3 and he tested on a 17R4, maybe that could explain this ~10°C difference.

 

Ah missed that. Still thats the 17inch version and just in Overwatch. Overwatch isnt really a demanding game by a longshot. In Destiny 2, Assassins Creed Origins etc i'm in the low 70s with about 140watts of power usage from the 1070GTX with non LM paste on the 15R3. Fully repadded.

 

Hi, I’ve just ordered all the stuff required to use thermal grizzly lm,
My question is before I do it is has any one seen what kind term affect it’s has on the chip and heat sink? Any thing to worry about and how often may I need to repaste. I see mixed opinions online weather to use it or not. But the results I’ve seem amazing, sorry if this had been covered but there’s 100s of pages to go through.

 

If the heatsink pressure is good etc it will last a loooong time. The heatsink will discolor to a silver color from the chemical process between the LM and copper but that has no real ill effect. If the heatsink pressure isnt great, then the LM can dry up on the chip die as well which is harder to remove.

 

If you have EVEN, complete and full contact pressure AND full insulation so that no oxygen can get inside the heatsink area, you should not have to repaste at all. This requires both an airtight seal and proper pressure.

A thick layer of Liquid metal on bare, completely exposed copper will have all of the gallium completely absorbed and will be hardened and oxidized fully within 2 weeks. I tested this on an open bench myself. With proper insulation, only a small portion of the galium will be absorbed, leaving enough liquid to last for many months.

 

Hello everyone. I have a (what is hopefully) a simple question that I couldn't find the answer to. I got an Alienware 17R4, with a 7700hq and 1070 recently. The build date is March of 2018, and my max temps using OCCT (both small and large datasets) are 68 in Optimus mode and 72 in dedicated Gpu mode, after 1 hour runs, with 21-22C ambient.

The highest temps I have seen are 74C after 2 hours of playing Kingdom Come with setting (almost) maxed out. I have already ordered Thermal Grizzly Kyonaut and the Fujipoly pads off of the links in the first post. My question is; did Dell change the thermal compound recently?

Do I still have to go through with the repaste or are my temps considered "acceptable" the way they are? I know the laptop is stock, because even though it's an R4, I bought it brand new, sealed, so chances of it being a used machine that someone did a repaste job on are slim to none.

Thank You for your time and huge thanks to the OP for the detailed instructions.

 

Is that only the GPU temp or also the CPU temps. If it is the CPU temp. Dont touch it for now would be my advice.

 

Thanks for the advice. It's the CPU temps. Gpu runs in the mid to high 50s. Do you think I should return the thermal paste or does it have a pretty long shelf life? If I were to not mess with it for now, it won't go bad in a couple of months right? Thanks again.

 

The thermal paste should last a couple of years when unopened and sealed in that black bag.

Those temps are excellent so I wouldnt worry about it for now.

 

Did the repaste with Thermal Grizzly, had a mini heart attack when I turned it on and the 1070 was nowhere to be found. Then I realized I disconnected the CMOS battery and the bios settings were wiped... lol.. Anyways, everything worked out, I got an average on 6C drop in dedicated GPU mode, I guess it might go down further once it cures. I did all the pads as well. One thing I noticed is fans don't spool up as much even with a 6C drop.

Thanks guys.

 

Kryonaut doesnt need curing, it tends to perform the best the first days. So keep an eye on those temps!

 

In that case it's a bit anticlimactic.. lol. Thanks for the heads up. I'm still glad I did the repaste, the old thermal paste was already brittle and cracking, despite the good performance and relatively new production date of the laptop.

 

Its moreike thermal putty so thats normal. But the performance formance isnt great indeed.

 

I've noticed lately my idle temps are up around 50degC, I was thinking maybe because increase in ambient temperature with heaters etc. being on now. I just hate hearing the fans when the laptop isn't under any load, the rear is already raised and I have an undervolt of 0.13v. Would a cooler master u3 help keep my temps down and be quiet or would I just be hearing the cooling pad fans instead? Or are there other things I can try first such as throttle stop profiles or disabling turbo boost to reduce CPU usage when not required?

Cheers!

Sent from my SM-G925I using Tapatalk

 

Turn of heaters and see if temps go back to how they were before. That will answer your question.

 

Yeh problem is I get home late and the missus has the apartment like an oven. Will try test soon in similar conditions as before when I was getting around 40degC for idle and no fans.

Sent from my SM-G925I using Tapatalk

 

So I just ran some tests in normal temperature room and yeh for some reason I'm still idling around 50degC with fans quiet but still audible. Yet when I bench test with Prime95 or even Heaven the Temps are mid 60s to low 70s.. Any ideas?

Sent from my SM-G925I using Tapatalk

 

Tried repasting myself with Conductonaut and new pads for the VRMs and chokes, here are the temps after 1h47m of gaming:




gtx1080 stock clocks (old vbios, never bothered flashing the new oc bios)
7820hk stock clocks, -0.130 undervolt
laptop raised for better airflow.

if I put my laptop flat on a hard surface, I get real bad temps. like all cores 90+ and gpu caps at 90 which is ridiculous...laptop should be raised further than that without the need for a stand.

 

Yes it's a silly design flaw. All they had to do with give the rear rubber feet greater height. Or maybe even some sort of "kickstand" mechanism.

The feet are so shallow, it struggles to get enoiugh airflow... all it does is make the system suck back up hot air when it is completely flat.

It's not possible to get sub 90c when flat on desk during full load for hours unless you undervolt and underclock a lot.

 

What are your CPU temps and fan speeds when idling?

Cheers

Sent from my SM-G925I using Tapatalk

 

Fan speed is always 100% (fan performance mode in BIOS) and idle temps are in the high 30s.

 

Fan performance mode means 100% fans.
You can enable CPU performance mode which spins fans when temps are >55-60C rather stock fans at 75C.