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

but what settings for overclock in Intel extreme tuning utility

 

The 17R4 has the biggest fans with the highest cfm ratings, however, the 15R3 isn't too far behind as I've had some units that were about on par with the 17R4's cooling system.

The config you have is one of my favorites as the 1070 is an excellent GPU that can be heavily OC'ed if you're lucky enough to land one with Samsung memory.

Edit: Oh I get what you were asking haha. Well, not all silicon are created equal so you'll just have to start small in XTU or TS and work your way up. There is no such thing as 1:1 systems...they are all different.

 

@iunlock Thank you so much

 

@iunlock
How can I know what memory brings the 1070 nvidia?
From what you say about samsung memory

 

use gpu z to see memory type

hynix ones are faulty and/or not oc as high as samsung

 

thanks!

 

Says (micron) is bad? :/

 

not bad, lower quality

don't pay same price for lower quality

 

Also Thermal Grizzly Conductonaut appears to be gone in the states. Is there any ETA on when sellers will get more? -nvm I was only checking big name re-sellers my mistake looks like the mod shops carry it.

 

I finally completed my repaste with CLLU and K5-Pro.
It was my first time working with either so it was a little sloppy. I probably used more CLLU than I needed to but I was erring on the side of too much over too little.

Unfortunately my results weren't spectacular; average CPU temperatures dropped by about 10C but GPU temps dropped by only 3C. Also, core0 is ~5C hotter than the other 3 when stress testing.
Running OCCT + Furmark averaged 69-73C CPU and 64C GPU (15R3/6700HQ/1070 with -180mV, 1cm tilt, +220/+764). OCCT alone stays under 60C and fans spin on low or medium.

The K5-Pro took 4 weeks to ship from Greece and was not very effective. I followed the instructions and slathered a ton on. I'm not sure if a small amount wouldn't have been better but the company's youtube videos suggest using a lot. Max GPU memory OC actually dropped from +805 to +764 over Dell's stock thermal pads.

I might attempt another repaste with smaller amounts of K5-Pro or different thermal pads altogether. Is there much more room for improvement?

 

Interesting to hear about the K5-Pro results. I assume you used CLLU on the gpu as well correct?

 

That's untrue. My Micron can OC to +500, maybe higher, not sure exactly because I undervolted to 0.800 . Samsung may be the "best" but it doesn't mean the other brands are bad. You even said Hynix was bad, which is COMPLETELY untrue, Hynix is one of the best, up there with Samsung. I am kinda doubt your knowledge you are spreading tbh. Just cuz Alienware needs to release a update for some notebooks so Micron is usable, doesn't make it bad, its pretty much labeled incompatible. Which is different than "bad" because incompatible just isn't usable, whereas bad just wouldn't overclock at all. Makes me want to checkout overclock setting to see how high I could take my Micron now

http://imgur.com/a/QW6dj

Wasn't going for a high score, was just showing that Micron is in fact, fine. At least on 7700HQ + 1070. 550 It does crash though, gonna try 500 + some core clock.
Changed my mind, not gonna change the core clock, too much of a pain for something I won't use

 

On my Omen 17 Samsung, can OC to 700 before getting negative hit in performance

 

But that doesn't mean Hynix or Micron is bad. For as long as I have been building my own PC's, videocards have been using different types of memory. It's not something that happens specifically with these laptops. It's just like silicon lottery. If a laptop performs perfectly out of the box, then the hardware is good. If it can go beyond the settings/clocks out of the box, then that is better. I'm not disagreeing that Samsung is better, I am just disagreeing with the comment that basically says "if you run GPU-Z and see Hynix or Micron, get the laptop replaced until you receive Samsung". After all, we aren't selecting Samsung memory chips, paying extra and receiving Hynix or Micron. It's just the way it works.

 

Yes there is more room for improvement.

Here is a 11 Hour OCCT test on a 15R3:

 

Samsung and SKHynix are among the best sitting at #1 and #2.

Micron is a toss up and the chips integrity is a lot weaker. Even their DDR5X used in the GTX 1080 are not that good compared to as if those GPU's had Samsung or SKHynix, which would be a lot stronger.

The truth is...OC'ing the GPU is a bit over rated as a lot will gain a little and most often times than not the price you pay for the extra heat can even bring about diminishing returns.

 

Yup, CLLU on both CPU and GPU.

I did slather the K5-Pro on super think as per the instructions:

I'll try repasting the K5-Pro with a smaller amount before tossing it away but worse performance than stock Dell thermal pads is disappointing.

[Edited for clarity]

 

When repasting with LM, it's not just a matter of a simple repaste and forgetting it. There is a lot more that goes into it like making sure the contact points are making good contact and this takes some work to balance the heat sink etc... If I were you I'd just go with Grizzly Kyronaut as that's pretty much fail proof.

 

I replied too quickly in my last post and it's badly worded.
To clarify, I slathered K5-Pro all over like in the manufacturer's youtube videos.

I used about a half rice grain sized amount of CLLU on the GPU and quarter on the CPU. It was spread as evenly as I could with the included brush. Temperature wise, the CPU stays between 53 and 60 running OCCT with fans on low or medium. I'm quite happy with the CPU temps but the GPU temps barely improved from stock.

I'll be opening the laptop back up and seeing if I can get better results. Thank you for making this guide btw. It has been a tremendous help.

 

Does anyone know if the fujipoly extreme xr-m pads will work on the 15R3. I have read that they are less flexible and less compressible than the other pads. Also how many pads are needed. The guide says 2X 1.0mm and 1X .5mm for the 17r4. Is this also what I should get for the 15R3

 

The GPU temps are very good even stock so there won't be a huge improvement there. This is normal so no worries.

As for the CPU, those are great temps. Have you unlocked your power limits? Any screen shots of hwinfo64?

 

I slapped loads of LM on when I did mine. Then I didn't get the results I wanted so I opened it up again and slapped loads more on too. I put about a grain and a half of rice on the GPU and about a grain on the CPU easy, then I put a thin layer on both heatsinks too.

Temps are now great but it's important to ensure heatsink sits square on both does also, this was the trickiest bit!

Btw iunlock did mention to use more LM on a laptop due to reduced pressure available I guess. It's why the tape is so important to catch any pump out. FYI when I opened my laptop for 2nd time I had no pumpout at all after a couple of weeks use. Also I used conductonaught which is less viscous than CLLU as well (so higher risk of pump out).

Sent from my SM-G935F using Tapatalk

 

are there any tips when adjusting the heatsink to reduce the core temp uneveness other than bending the tripod tension arm?

 

Perhaps this is better suited in this thread:

How come the OC in the BIOS seem to cause so much higher temps than doing the same OC in XTU? When using XTU, my temps seem to be about 10 degrees lower at the same speeds. But I don't really know what the options in the BIOS actually do....besides frying the CPU and causing throttling that is.

 

I repasted the K5-Pro and added a bit more CLLU. This got rid of the 5C differential on core0 but there was no improvement for the GPU.
I didn't realize power limits played much of a factor for a locked CPU like my 6700HQ? The CPU ran at 3.1GHz on all 4 cores with no throttling.

Ran OCCT for an hour with -180mV, 20C ambient and 1cm tilt.

 

For some reason I had thought you had the 6820hk... Yawwnn Long day lol.

Anyhow, Great Temps. You could enable performance mode in bios and although it will show a higher limit your CPU will still be capped. It should change your fan curves though.

::iunlock::

 

@iunlock, @Pete Light and @Harmful : Do you have a picture showing how much amount of TIM you've applied for CPU and GPU?

 

@iunlock @Pete Light you both mention the heatsink being much better when flat, (I know that's always best) but what methods did yous use to achieve 'flat' to your standards? I received my copper shims (haven't used them yet), but I think it's more likely I need to use more LM as opposed to making sure it lays flat.

Also... dumb question. Does anyone know where to get The same/similar 2 screws that go on that back near the charger port, HDMI, DP, ethernet, etc? Stupidly, I picked one up and screwed it into the bottom. Prettttty sure it was too big for the threads because by the time I noticed it was the wrong screw and it got tight, it also just doesn't want to come out. So I may have to cut it or damage it to get it out haha.

 

See pict first page!!

 

@Papusan: Same amount for CPU also?

 

Ps. CPU die is smaller. Hence less paste. Better to start with less and then apply more if needed.

 

No because it's smaller
I put about a third less

Sent from my SM-G935F using Tapatalk

 

Should I apply one-third pea method or line method for CPU?

 

If you talk about liquid metal, it doesn't matter. You have to paint it on die anyway. Not the same as conventional thermal paste.

 

Using traditonal paste or liquid metal?

With traditional paste you can over do it and be just fine, in fact even better than risking of not putting enough.

With liquid metal, it has a smaller tolerance level to where you don't want to over do it, but at the same time don't under do it or your Temps will be worse than before.

Tip: It's best to not fully reassemble the machine right after you repaste... After your mobo is back in, secure it with the 4x silver screws and 4x silver fan screws then prop it up safely and turn it on to test the Temps before putting it all back together.

::iunlock::

 

@Papusan and @iunlock: Thanks for the tip. I'm not using metal paste rather traditional paste as the former isn't available. The fact is I didn't apply enough paste causing high temps at full load. But in idle mode its 28C-38C depending upon ambient temps.

 

I think this got buried so I will try again. Will the fujipoly xr m pads work for the 15r3. I see the xr e are used in the guide but the xr m are more conductive but I have read they are less flexible and compressible. Does anyone know if they will still work. Also the guide says 2x 1mm and 1x .5mm for the 17r4. Is this also what I should get for the 15r3? Any advice would be appreciated.

 

I'm just about to pass the 6 Hour mark on Fire Strike Combined Test Looped for 6 Hours. This is with Grizzly Kyronaut at 40x and the temps are respectable for it being traditional paste.

I'll post some screenies when it's finished....but I'm asking myself, when will that be? LOL ...as the temps seem to have stabilized a long time ago so really no need to go longer eh.

I"ll stop it at the 6 Hour mark...,

 

just to clarify, when you specify 3x - 1.0mm (100x15) with regard to the thermal pads...that means I need 3 sheets of 100x15 in order to cut enough pads...correct?

 

Use the fujipoly 11.0's. I'll be updating the OP shortly. No more 17.0's.... It dries up and flakes like chalk.

Correct.

::iunlock::

 

Never liked these 17.0's. They are too stiff. No flex
See also Thermal Paste Roundup

 

Yea...they did perform well, but the dry out effect is no bueno. It's a one time use pad for sure and not worth double the price to that of 11.0's.

 

This is interesting. I don't know the price for 14W/mk vs. the other.

 

Ah yes the ones from Germany. Kind of hard to get a hold of in the States, but I'm working on getting some straight from the source.

 

I live in Germany and always prefered ac over fuji. Cheaper and better to use

 

So i repasted my 15 R3 with LM Conductonaut and i am disappointed. I am not sure if i didnt put enough Paste on CPU and GPU but my CPU gets 100C hot when gaming. Undervolted -150 and it still gets up to 94 on Core 0. Cores temps are sometimes 10C different .

This was my first time using LM. Do i need to apply quite a bit on Die and Heatsink? Need some advise before i take it all apart agin

 

I would resort to using Grizzly Kyronaut if this is your first time. Your temps should not be that high. LM is tricky to work with...

Before: Stock


After: w/ Kyronaut


Great temps can be had even with Kyronaut.

 

I got the paste now, dont really wanna go and buy Kyronaut. Tricky how? Should i take my laptop apart and then add Paste?

Does LM need to settle like regular paste? Like heat/cooling cycle?

 

Hey guys....precursor?

If this is any indication of what we could expect with the new AW17R4 / 7820HK / GTX 1080 on the new vbios....

This are some respectable numbers....I'm confident to break 19K's....just getting started.

Asus G701VI / 7820HK / GTX 1080


228W before I had noticed that the CPU fan power cable wasn't even plugged in!


Explanation:


---

How will the new AW 17R4 / 7820HK / GTX 1080 do? Will it be able to stand up against the new Asus G701VI / Kaby Lake? We'll know shortly, but if it's anything close to what we're seeing with the Asus....this should be an exciting ride...

 

LM does not require any cure time. The nice thing with LM is that you could simply just add some more on top of what is there already. Be careful and good luck!