[CPU + GPU Temperatures + Benchmarks] - XPS 15 [9560] Kaby Lake

Well, undervolting is the best way to lower temperatures. On average, my temps drop by about 1C per 10mv undervolt. And by comparison, my CPU (7700HQ) is always below 0.95v even when fully under load, while your is still at 1.032v, which is still a pretty high voltage and will be a large source of heat. It's weird that you can only undervolt by 120mv though, most 6700HQs can get 150mv+.

I am sadly no expert on liquid metal, so I cannot give you advice on whether this is a good appliance. The best guess I can give you is that it seems like it is too much liquid metal, especially if you compare it to images on Google: https://www.google.com/search?q=liq...UIDygC&biw=2197&bih=1092#imgrc=Xo4QMTATf1AA_M:

Also, I believe there is a BIOS hack somewhere here on the forums that allows you to undervolt the dGPU in the 9550. That should further help with managing temps.

However, it may just be possible that you have especially bad binned versions of your chips and this is simply the max attainable.

 

oh thanks for the feedback!

hmmm yea, I may have been getting bad binned versions. Unfortunately, I swapped out 3 motherboards and all 3 produces around the same amount of undervolt. I killed 2 previously in 2 years and 6 months and swapped them out thanks to Dell warranty because I used them till they overheated.

Also, I need to perfect the liquid metal application on my own.

I will also dig into the forum to undervolt the GPU, thanks!

 

> Some of it actually spilled out a little on to the electrical tape, hence I think I may have put too much Liquid Metal.

Maybe indeed.
I guess you could check if your heatsink is aligned perfectly and the gap between it and the chips is as thin as possible all over, with as much metal-to-metal contact as possible.
You can try this by applying tiny dots of regular paste near all chip cornes, attach and then remove the heatsink, see if all dots spread nicely.

But your measurements don't show thermal nor power limit throttling throttling. And I don't think these temperatures are problematic. Do you have any performance issue?
Have you padded the VRMs?

 

I hope you're not killing these motherboards with liquid metal and then getting them repaired under warranty...

 

Reply to _sem_

I haven't tried it, I will go try it and then let you know! But I do know what for the previous few heatsinks they sent over, some came warped or they had too much distance between the silicon die and the heatsink. But this current one that I have, I applied some amount of pressure on them before closing the lid of my laptop so that the contact between the silicon die and the heatsink is better.

Yes, I have padded the VRMs, I want to replicate iunlock's mod for the 5510/9550, but then the capacitors in between the VRMs, and the capacitors are taller than the VRMs, hence it makes me scared about putting heatsinks on the VRMs cus I don't wanna fry the motherboard with heatsinks.

Reply to custom90gt

I didn't apply any liquid metal to the previous motherboards that I had. I only started trying out liquid metal with this motherboard that Dell had sent over to me for replacement.
I killed the previous few motherboards because of high intensive CPU + GPU usage for over too long periods. And since heat kills chips due to electromigration and other sort of phenomenon under normal usage, hence I decided to liquid metal the CPU and GPU on this new motherboard.

EDIT: I only gotten this new motherboard 3 weeks ago, and bought the liquid metal syringes about a week back. And because I wasn't getting the temps that you guys were getting, that's why I came to this forum. Also I watched LinusTechTips (he made a mistake? He was supposed to use the smaller metal needle-like thing that came with the Conductonaut, not the black plastic thing used to suck up liquid metal in the event too much came out, but I think he corrected it??) and a few other channel on how to liquid metal a laptop.

 

> But I do know what for the previous few heatsinks they sent over, some came warped or they had too much distance between the silicon die and the heatsink. But this current one that I have, I applied some amount of pressure on them before closing the lid of my laptop so that the contact between the silicon die and the heatsink is better.

Also mind VRAM pad thickness.

> hence it makes me scared about putting heatsinks on the VRMs cus I don't wanna fry the motherboard with heatsinks.

But you do play with LM, hehe.
Mind spilt LM is "corrosive" (gallium migrates into other metals, makes some like Al brittle).
I guess taking care of the VRM heat is more important in XPS 15 with high loads than LM. The CPU and GPU don't die at these temperatures. But under load the heat tends to accumulate in the VRM area.
http://forum.notebookreview.com/thr...thermal-insanity-as-seen-through-flir.826949/
Have you uninstalled Intel DPTF? Unfortunately, XPS 15 isn't meant to run without the throttling.

 

Regarding the VRAM thermal pads, I changed it out for squishier thermal pads of 0.5mm
I had actually padded the Voltage Regulator Modules with Fujipoly 11.0 W/mK Thermal Pads
But anyway I still hit Power Limit Throttling, after about 5 minutes of stressing, the screenshot I took is before the laptop hit Power Limit Throttling. Padding the VRMs will result me in 84 celsius in the VRM area when both CPU and GPU are stressed at the same time.

I know Liquid Metal reacts with Aluminum, Copper and even Nickel. However, Liquid Metal reacts slower with Nickel and Copper.
With regards to Liquid Metal and Copper:
The higher the temperature, the more kinetic energy the molecules of the liquid metal has, will result in more frequent higher energy collision between the liquid metal and copper and hence, the reaction we see after applying liquid metal. But even then it will take probably at least 6 months to see the newly formed copper-gallium alloy and the new alloy wouldn't affect thermal conductivity too much.

I have Intel DPTF installed. Unfortunately, I live in a tropical location with temperatures averaging 25 celsius to 37 celsius on certain days. So I am quite afraid to uninstall DPTF as I have read that it could potentially bust the hardware of the laptop if it overheats especially with the Precision 5510 / XPS 9550. But I will take a shot at it.

With regards to applying tiny dots of regular paste to the heatsink and then putting it back, the paste spread out pretty evenly.

 

> I had actually padded the Voltage Regulator Modules with Fujipoly 11.0 W/mK Thermal Pads
But anyway I still hit Power Limit Throttling, after about 5 minutes of stressing, the screenshot I took is before the laptop hit Power Limit Throttling.

It has been reported before that hi-grade pads for the VRM mosfets
- still don't take away enough heat (this could likely be worked around using metal heat conductors, best pads conduct heat 1-2 orders of magnitude worse)
- tend to make things even worse in the long run unless using a vented cooler plate, because the backplate heats up, in turn intake air heats up... Even with a vented cooler plate it may be tricky to remove heat fast enough.
Hence iunlock adds internal "fin surface" for the VRMs and redirects some cooling air over there, to take the VRM heat out via convection.
Laptops with more elaborate cooling tend to have heat-spreaders covering the VRM mosfets, or even extra heatpipes for them. So their heat exits via the cooling fins. There were ideas for beefed-up heat sinks for the XPS 15 at dell reddit, but no well-documented success story so far.

> With regards to Liquid Metal and Copper: ...
I hear Cu heatsinks are not problematic, except that staining is visible, so warranty may be denied.
CLU users reported performance degradation and LM solidification after some months of use, probably because gallium migrates into copper and the galinstan alloy in the gap is no longer at the right percentages.
Other metals might get affected adversely due to spills.

> So I am quite afraid to uninstall DPTF as I have read that it could potentially bust the hardware of the laptop if it overheats especially with the Precision 5510 / XPS 9550. But I will take a shot at it.
Don't, considering you've fried mobos before

> With regards to applying tiny dots of regular paste to the heatsink and then putting it back, the paste spread out pretty evenly.
I think that if the gap is as thin as possible, LM doesn't make so much difference. Except if you're at the very edge of direct thermal throttling at the CPU/GPU, which doesn't seem to be your case. Think about the heat flux from the heat sources towards the ambient, not just temperatures. You must get the heat flow out of the laptop a bit better, while not hitting the undisclosed throttling temperatures anywhere.

 

I recently bought heatsink fins similar to what iunlock had, but however because of the capacitors that are taller and are in between the voltage regulator modules, are there

1) any other methods that I can use to dissipate the heat?
2) smaller heatsink fins that I could install such that I don't short the capacitors on the motherboard. I don't want to see my motherboard dead because sparks flew or because of heat.

I know which Voltage Regulators to put heatsinks on, I just don't know how to go about doing it because of protruding capacitors.

I recently read from iunlock that stacking thermal pads is a bad way to conduct heat away. Which was why I was initially against the idea of padding the VRMs, and then I went to buy same exact heatsinks that iunlock used in his guide (can be found here: https://www.adafruit.com/product/1515) but then i realised that the capacitors and the heatsink fins were in contact (while the laptop was still turned off, grounded, no power supply connected), then I thought to myself, maybe I should remove the heatsinks, just in case something goes wrong.

1 possibility to prevent it from shorting would be to spray the side of the motherboard with the VRMs a layer of clear enamel nail polish, just like what I did on the GPU capacitors before liquid metalling my system. What do you think about the nail polish method?

or if i can find an alternative heatsink fin to the ones he used in his guide, those that are way way smaller.

Thanks so much for all the info and guidance!!

 

Hi. Sry I don't know where I can ask my question. I buy aw 17 r1 and change bios over clocking and now the laptop show black screen. What can I do ? I open the case cover and pull up CMOS battery and laptop battery and push them again , but nothing happens and black screen come again. Unfortunately the ram is on slot 1 or 2 not 3 or 4. How can I reset bios without dissemble PALM rest?plz help me I just bought it yesterday. Plzzzzz

 

> because of the capacitors that are taller and are in between the voltage regulator modules

I guess these are coils?

> smaller heatsink fins that I could install such that I don't short the capacitors on the motherboard

I assume their surface is not electrically conductive, you can check. But it is good to leave some space for airflow. Mind air for iunlock's VRM cooling should enter via the gap above the heatpipes, leave via the small central grille, and meanwhile go past the heatsinks and the copper foil.

You can use a file or sand paper to reduce the size of the heatsinks, if you think they are too big. It will be difficult to find something smaller. AFAIK small copper heatsinks for mosfets with nail-like fins exist, but you'd have to trim those too.

 

mmmm, the huge silver coloured ones are the chokes,

I am not referring to the chokes, but I will take a picture for you tomorrow. Sorry for replying late, had been really busy.

 

Picture: https://1drv.ms/u/s!AvIsJ-c9AzQV9D7L3XyyeEvmZOW6

here the capacitors I marked out around the Voltage Regulator Modules


sorry i have been really busy

 

Yep these surely have metal contacts one shouldn't short to sth.
As said you can cut the heatsinks to size. And you must be extra careful to fix them properly, so they don't cause a short upon impact. But using insulation also reduces heat flow.

 

Don't let anything metal near those capacitators. Give some clearance. If metal heatsinks come loose, there are plenty of places to short out.

There are adhesives for metal heatsinks but they will likely void your warranty, unless you can remove the DIY mods leaving zero trace.

 

I was just thinking about putting a thin layer of "thermal tape" under the heatsink, i tried that but to no avail, suffered heavy throttling.

Edit: if you do notice, the GPU VRMs are blocked by the tall R22 chokes. and the CPU ones are also being blocked. :/ maybe that's why I was suffering from heavy throttling.

What can I do to resolve throttling? hmmmm

 

Also, I realise that once the laptop enters Power Limit Throttling, 0.8GHz CPU speed will be engaged, VRMs cool to 64 celsius and cannot get out of Power Limit Throttling, it will just lag forever.

 

Does anyone else's 9560 get really hot near the headphone jack when charging a low battery? If I do a full charge cycle when the battery is low 20-40% the area just below the headphone jack will slowly heat up to 81C for a few minutes and then cool back down to 50C. I suspect I need to add some thermal modifications to one of the components in this area:



Perhaps I could take a piece of the copper tape I'm not using with a heat sink and thermal tape spread the heat out over the area.

 

Yes there seems to be a charging circuit there. I recall there was a thermal image, maybe in that 9560 thermal insanity thread. There were a few reports of excessive heat, maybe after battery replacement. Someone mentioned padding to the backplate. You could also consider thermally mating it with the fan housing.

 

I didn't cover the entire mosfets with thermal pads but I did recognize they needed padded.... Guess I did a half job?

 

@Philaphlous I was just looking at your flickr album (thanks for documenting all that) and was curious what your final configuration you settled on for all the thermal padding. Do you still have thermal pads on top of the head sink pipes above the CPU and GPU as seen in this photo?: https://www.flickr.com/photos/30149337@N04/32283599404/in/album-72157680615207736/

Also, I noticed you taped some of the air gaps over the fans, did you just use electrical tape for that or is there a special brand/material that should be used?

I just repasted my 9560 with Grizzly Kyronaut and did the same configuration as you have above the heat sink of small cutout 3x stacked pads over the the mosfets, but just those 4 areas padded, nothing else. It's been a huge improvement for me so far, also using ThrottleStop with -120mV undervolt on CPU, CPU Cache, IntelGPU. I'm not reaching the benchmarks that many others have posted, I get 2787 in Heaven for example, you had over 3000, so I'm looking at copying the extra pads you added as well as that tape.

 

I do the tape too. Use something thermally rated above 100C just to be safe. Regular electrical tape would be around 60 C if I'm remembering correctly.

Also, make sure to do in a way so no exposed sticky side exists where the gaps are

EDIT so I don't double post.

@iunlock and @custom90gt and anyone else who's attempted iunlock's mod:

How well does it hold up overtime with lots of traveling? I commute a lot, sometimes by train, sometimes by plane. Because of how much my XPS is in motion and packed into/out of bags, I'm wondering how secure the aluminum heatsinks are when used with the adhesive they come with or with thermal tape.

 

I use Silverbead thermal glue which seems to be the same stuff that I often see in PSU.
It stays a tad soft, similar to silicone, but is really really strong. I use it with 8x12x3 mm heat sinks and
a much simpler version of the "air redirect over VRM" mod. I look that I can post some pictures later.
My 9550 is also traveling a lot and I have no fear that the thermal glue will fail.

edit: My VRM do hit 90°C at CPU+GPU at 100% though, but I am perfectly fine with that as
it was hitting 105°C and thermal throttle before.

edit2: Sadly I am unable to find a fitting screwdriver, so here are some older pictures.
The heatsinks are different, but the rest is the exact same.
They were a little too tall so I replaced them.

edit3: I just remembered that I lifted the right (in the picture) Fan with another strip of
one of those thermal pads so that more air goes above the fins and to the VRM.

edit4: I personally would never trust thermal tape. My sample size is small, I only tried two
different products, but they are really flimsy and don't hold very good for small heatsinks.
Even though I properly cleaned with alcohol.

 

I have a Dell Precision 5510, similar to XPS 15 9550, except for the CPU and GPU configuration, but iunlock's method works really well.

Personally, I had been using iunlock's method for a few months and travelling is not a big problem. The heat sinks won't fall off as the adhesives that are really strong.

I got my heatsinks from here: https://www.adafruit.com/product/1515

 

hello everyone,

need your help, guys.

my xps 9560 started to throttle hard:
IA: Package-Level RAPL/PBM PL1 - yes
IA: Package-Level RAPL/PBM PL2,PL3 - yes
IA: Electrical Design Point/Other (ICCmax,PL4,SVID,DDR RAPL) - time to time yes

GT: Max VR Voltage, ICCmax, PL4 - yes
GT: Inefficient Operation - yes

RING: Max VR Voltage, ICCmax, PL4 - yes

GPU: Performance Limit - Utilization - yes

While my temps are pretty low:
CPU - 89 max, GPU - 77 max, rest temps all below 75-80.


Any idea what can cause this throttling? Started over month ago, before everything was good and swift.

 

I'm not sure why all GTX 1050 graph tutorials have the mv at the bottom startin gat 800 but mine is 700 :-S

 

Hey all, I have read a bit of info saying if you also 3 stack pad the PCH Chip and it can help reduce temps a lot. I am trying to firstly identify which one this is exactly and secondly if it is even worth doing?

 

Anyone else notice any drastic improvements on gpu overclocking over time I'm on the 450.99 developer preview drivers with the latest 2004 windows 10 update and I'm already 100mhz over what I previously though was stable on the memory overclock... Still benchmarking and stability testing but I've never been able to get over +240 on the memory and I'm at +330 and climbing... Very odd... I know my memory was binned a little lower than what some people have been getting but maybe it was driver/system related?

What's the highest memory overclock you guys are seeing on your 1050?

wow...now I'm up to +500... I've never seen this before...what is up? I'd instantly crash anything over 300 before...

 

Benchmarks aren't much better with the huge increase in memory overclock... For superstition at high... I maybe get 30 more points going from +200 to +450. I wonder if the new drivers are heavy on error correction and that's why I'm not seeing much of an increase in benchmarks...

 

From what I read, you don't see much gain from memory overclocking on any 10XX series or later GPUs.

 

Hi all!

I own a Precision M5510 (i7-6820HQ, Quadro M1000M, 16GB RAM, 4K IGZO).. Been using this laptop mainly as a mobile photo/video editing rig and has served me very very well.

Prior to this whole reading of this awesome thread and the input of everyone here.. my M5510 suffered a PCH failure (as prescribed by the laptop repairmen) so i took the liberty of ordering off a used motherboard and did the transplant on my own.. Powered on and it was heating badly... so i thought of replacing the thermal paste (as any first thermal related solution would be). yeah... hardened thermal paste before the work..


Here's what i've done so far on my revived M5510...
<<Physical Mod>>
- TG Conductonaut on CPU , TG Kyronaut on dGPU (im afraid of spilling LM on the small chiplets on the board..)
- Lightly polished and cleaned off copper heatsink with metal polish abrasives(was badly tarnished till green..)
- Kapton taped the opening between heatsink and fan.

<<Software Mod>>
- CPU Core/ CPU Cache ... (AC -170mv, Batt -140mv)
- iGPU ... -50mv
- set nvidia max fps 60
- EPP 64 on AC, 255 on batt


Never had much chance to test a video encoding and much... so i only used Unigine Heaven and ran it for a good 1hr for each ThrottleStop tweak..

Before:
- CPU 99deg (throttling nonstop) GPU 90deg

After:
- CPU 70max, gpu 75max...


As for games, thanks to the info of limiting max FPS to 60 (display is 60hz anyway...) most of my games dont hit both cpu/gpu at 60.. and i mainly play World of Warships.. L4D2.. just to kill time..

currently procuring all the cooling stuff that @iunlock has so painstakingly documented and further bring temps down...


I have a question to anyone's reading and have the answer for... i cant seem to activate the voltage curve for MSI AB (not sure if my GPU is not supported)

Also, i do read someone mentioning that the heating problem on the left side of the laptop whenever the charging cable is inserted... anyone has a ultimatum on that ..?

i apologize in advance if i may have missed out any of above said solutions to my questions....

 

How did you change the PCH?

 

haha i didnt.. bought a used motherboard from ebay for about 400usd..

ouch i know.. i love the display on this thing and getting a new laptop is so much more expensive haha.