XPS 15 9550 temperature observations (undervolt + repaste)

Well I decided to save the homework for tomorrow morn and went ahead and repasted the CPU and GPU but don't think it helped much, if any as it is still throttling. It's crazy to think that all that little area in the back needs is a little bit of airflow. Just lightly blowing on it will drop that Ambient/Aida64-DIMM temp 20 degrees.

Just for ****s I decided to borrow a thermal pad from something just above the battery cant remember what it was, and place it over that back area as well as the heatsink (since I didnt want to put the back on yet) and since that back area temp is higher than my heatsink I figured it might help a little bit. It prevented temp from reaching 103 degrees - which seems to be the throttling temp, but instead stayed right around 100. Still way too hot.

Oh I forgot to mention, this has been with TurboBoost disabled , and a -140mv offset using ThrottleStop. So I guess I might buy a few thermal pads and see what happens. If I knew returning the thing would solve the problem I would, but the likelihood of the new one doing the same thing is pretty high I would say. I will also try to set the CPU clock max performance to 80% through the power options.

 

A few comments:

1. I see you are disabling Turbo via BIOS. Make sure your did not click the "disable turbo" box in ThrottleStop as it was made for older processors.

2. With newer chips, undervolt may require that you set the same offset for both "CPU core" & "CPU cache". Otherwise the undervolt will not work.

3. What are you running to get those temps? Games? Benchmarks? Don't tell me FurMark

4. You may have a defective unit. There are a couple of posts here of people with 9550 that had thermal defects in their computers and returned them. If I recall correctly, replacements worked better. Do a search here.

 

Well, heres what I set:

Max CPU: 80%
TurboBoost disabled via bios (not checked in ThrottleStop)
CPU and Cache -140mV with ThrottleStop
Repasted GPU + CPU
Side Note: My results were quite different whether or not I used my laptop cooler
Benchmark: Ive been just using Unigine Valley. I did play Overwatch briefly last night and it would start throttling only like a couple minutes into the first match.

Without Cooler:
Aida64 Graph: http://imageshack.com/a/img923/3104/LxAcJA.png
*Ambient temperature (this is new) + CPU temp stable, but GPU Thermal throttles a bunch
MSI Afterburner graphs:
*Top half: http://imageshack.com/a/img922/9948/fK0SHt.jpg
*Bottom half: http://imageshack.com/a/img924/7831/C0jbTC.jpg


With Cooler:
Aida64 Graph: http://imageshack.com/a/img924/9085/ko44Jf.png
Just another trial - http://imageshack.com/a/img923/6103/PTKEAh.png
*CPU + GPU Temps stable, but Ambient temperature slowly elevates until it throttles (much greater amplitude than GPU thermal throttling)
MSI Afterburner graphs: will include this in just a moment



Well, in either scenario I have throttling one way or another. Without the cooler the GPU throttles almost constantly although at a much lesser extent (drops to like 1400+ MHZ, instead of the 800-1200 with the VRM throttle). With the cooler It takes much longer to throttle, but when it does it hits a lot harder. I think its pretty dumb I have to downregulate all of this just to prevent any sort of throttling. Thanks for the help guys, don't think theres much that can be changed except possibly try some thermal pads or exchange for another.

I might start a thread for 9560 owners to get an idea of how prevalent this is - if you guys think its a good idea that is, im new to the forum.

 

Probably from the SSD?
The thermal pad is not meant to be a heatsink, rather a thin layer that conducts heat to the heatsink. Here, the idea is that the bottom cover starts serving as the heatsink for the VRMs.
Several others have reported that padding the heatpipes is not a good idea, the bottom gets too hot and things get worse.

 

Yeah SSD, it troubles me that you are inside a laptop messing with things and do not know what an SSD is
Just having something to spread the heat can help such as the copper sticker on the 960 pro that helps remove heat as it increases the area for dissipation.

It still needs putting back as it is in place for a reason!

 

The one I got on SSD was too thin to work I guess, perfect for VRM area experiments. But I replaced it with a thicker one on the SSD.

Btw, I understand there is a conceptually similar copper sticker on the inside of the bottom alu cover under the black foil?
http://forum.notebookreview.com/thr...ty-display-2015.776840/page-277#post-10143762
But that thermal image seems to show it doesn't help so much. Might a thicker one help spread the heat better, combined with some pads to the VRM chokes? If there is space... I saw one choke at the edge already pushes against the material on the bottom cover (no space for a pad on its top). And one should be careful towards the vent area not to wreck the sealing of the exhaust and let hot air leak to the VRM area instead of going out through the fins. In fact we might as well check if this happens to be already going on and causing the issue.
http://forum.notebookreview.com/thr...0-owners-lounge.784473/page-102#post-10347023

 

Has anyone removed the fan plastic mesh grill from the blower section?


Looks like the round holes really prevent decent airflow. Obviously it wouldn't be noticeable looking at the laptop if the mesh is removed or not. I did this on my old XPS 14 with a good result. Wondering if anyone has had a positive and noticeable outcome from this. I would imagine it would also reduce the volume since the air isn't being as restricted as much.

 

I like the idea assuming you could remove it without destroying the case. Sadly I don't have any time to try it myself...

 

I don't think that will work on the 9550.

On my laptop, the two radiators to the left and right side of the back grille have no round holed panel encumbering exhaust.

The round holed panel is only at the center of that grille; behind that holed panel is a foamy pad "insulating" the VRM and preventing airflow in that area.

But feel free to see if that is revised on your system.

 

Ive been doing quite a big of testing, while also trying to understand the results. One thing is for certain, and I'm sure many people already know this, but a laptop cooler seems to completely change the airflow dynamics for the worse.


With a laptop cooler - https://www.amazon.com/Zalman-Minim...coding=UTF8&psc=1&refRID=KP9GJCSXT2RRC66MH6YW
VRM temp: rises very quickly to ~100-103 C until power voltage is down regulated.
CPU temp: fine
GPU temp: fine


Without my laptop cooler:
VRM temp: Maxes at 87 and does not increase (13-16 degree difference).
CPU temps: fine (Max CPU temp I saw was like 77 or 78)
GPU: thermal throttles quite regularly at 77-78 (max GPU temp I've seen is 78) (isn't this a little bit low to thermal throttle?).

Unfortunately fairly soon after my GPU starts throttling, my PL1 drops to 7Watts, which at that point my system seems to be completely screwed for the remainder. Seems to take ages a to recover.

Why is occurring? Is it due to the amount/rate that the GPU consistently throttling which then the system decides to prevent it and power limits? Or is it simply due to using too my power? I feel as it is has to be thermal related, or else why wouldn't the PL1 drop to 7 right away during high load.

 

I thought about it when lifting the tape covering about three quarters of it on the inside. But already lifting that tape should make a notable difference in the open cross-section, but it didn't seem to make much difference in PL throttling using Prime95 and Heaven at once. Also, that grille looks much more "transparent" if you're watching it straight, not from such angle. Lifting the rear (not letting the exhaust air go below) made a more obvious difference, resulted in higher CPU freqs on average (while GPU was not limited mostly).
I guess the difference is little most likely because the vents don't seem to be able to suck much from there anyway, the air passages under the bottom cover seem to be narrow. Closing the grille completely didn't have any effect, but blowing from the outside did. I tried to pad the VRM chokes as to avoid blocking the air from going past as much as possible, now I guess it doesn't matter much if the whole VRM area is stuffed with pads, as long as they don't touch the heatpipes.

Maybe someone can repeat the experiment with another GPU hog which produces a more stable load on the GPU, because the changing of the scenes in Heaven introduces changes of GPU load and makes it more difficult to assess the difference.

 

GPU itself should throttle a bit over 90. If lower, sth else must be triggering the throttling. Also, the GPU throttling I've seen before repasting was quick, accurate and stable, no large and slow zigzags in the temperatures, just some noise.

Also notice that there is a strong correlation between some temperatures and the frequencies. So if the CPU and GPU can run at higher frequencies without throttling, some temperatures will rise higher than what they would be with throttling. With closed-loop control and multivariable systems it is not always quickly obvious what is the cause and what the consequence, particularly if you don't have the diagram of the control scheme.

 

How much did you overclock 1050? 1700 Mhz?

 

Its currently at stock speeds, which according to afterburner between 1708 and 1746 Mhz @ ~1.05v or somewhere right in there.

Just wanted to add that, after running valley in a cold environment (highest temp was 65 C), had no issues with PL1 drop, indicating its thermal related. My highest temps while inside are VRM: 87 and GPU 78 (throttle temp) - which unfortunately despite being sub 90, trigger the PL1 down regulation.

 

wow.. thats quite a bit higher then offical 1050m clock speeds (which should have been 1354 base and 1493 boost at 7000 MHz memory, which gives 25% boost compering to 960m, considering that Pascal and Maxwell are mostly the same architechture just on the different tech process). That should give a 1050m about 40% boost next to 960 1200 mhz boost speed.

 

When I get my XPS 15 9560 I'll be making a diagram of all the VRM area and observed temperatures on multiple loads, GPU and CPU. This will hopefully give us all a good idea of where the heat is coming from. I don't have a FLIR camera so I'll be using my infrared thermal gun and recording temps all around... HOPEFULLY I'll be able to use thermal pads on the VRM areas that will reduce temps from throttling...

 

How do you know it is the GPU temperature? With HWinfo64 or Aida you can see a bunch of different temperature measurements. I guess some of them trigger throttling. The misnamed "DIMM" temperature (sensor next to the grille) was already mentioned, I guess there must be at least one more.

Though, if the issue is generally that the VRM area is overheating, it might be a better idea to focus on locating heat sources and on how to take the heat away a bit more efficiently towards the cooler areas around the vents.

Is there a pic of this copper strip hidden below the black foil on the case bottom? We might want to replace it with a thicker/larger one?
http://forum.notebookreview.com/attachments/dell-xps-15-9550-aluminum-bottom-inside-jpg.129486

 

Actually, the whole black foil is painted copper.
http://i.imgur.com/eAFtMJf.jpg

 

Philaphlous,

There are a few 9550 images to get you started including:

1. Thermal picture of 9550 by Bokeh (actually similar Precision model. I wouln't run FurMark)

http://forum.notebookreview.com/thr...0-owners-lounge.784473/page-102#post-10347065

2. Clear 9550 VRM photo by GoNz0

http://forum.notebookreview.com/thr...observations-undervolt-repaste.785963/page-27

3. Draft diagram of the 9550 VRM with part numbers. It is not all correct but could help you. If you can't access the .jpg attached to post below try a different browser. If that fails, contact the person who posted it

http://forum.notebookreview.com/thr...ndervolt-repaste.785963/page-48#post-10333465

 

That's great info! Thanks!


Could anyone answer for me the thickness between the aluminum frame to the VRM? Wondering how thick of a thermal pad I need to get...2-4mm?

 

A few people tried 1-1.5mm thermal pads on the 9550's chokes per the following posts.

I think the VRM photos for the 9560 & 9550 have different choke part numbers so the sizes might be different.

Also, you might want to use soft fluffy thermal pads lower performance pads vs. the super dense/hard high-performance thermal pads.

- May reduce probability of damaging the delicate choke-board adhesion
- Requires less precise thickness estimation
- I found that high-performance pads made the keyboard super hot and potentially REDUCE laptop performance (see the posts for more)

 

Pressing,

Thanks for the info! That's great to know! Wow...sounds like there's some incredible issues with thermal throttling... surprising with such great chips in the laptop..I guess its really pushing the boundary of how much power it can actually draw... I've got a bunch of soft silicone 4 w/mK 2mm thermal pads already so I might just play around with those a little... Best bet sounds like CPU undervolting is really going to help the thermal issue. Is the VRM primarily supplying voltage to the CPU or GPU that causes the throttling?

 

Some simple user tweaks can really boost performance of the XPS; this is a helpful thread to evaluate what worked for others and try some new things

Not many laptops this small are running 45W CPUs and dedicated GPUs. If you can keep the VRM temperatures low, it can provide max power to CPU & GPU. As the VRM heats up, power supplied seems to drop off a cliff (mosfets get inefficient)

See how that works on the chokes. Note the warnings in earlier posts (keyboard gets uncomfortably hot, bottom case gets too hot and superheats intake air, spiraling all thermals out of control). A few guys said they got a good balance for gaming with low W/mk pads

Undervolting dropped temps by over 10*C and the repaste another 5*C on my laptop but the benefits are even better as you remove bottlenecks. Since the whole system runs cooler, I think the VRM is also cooler (but I can't remember). What I do remember is that once the CPU & GPU temps were under control, the VRM was the bottleneck...

Don't know.

- The CPU runs 45 Watts and the GPU maybe a bit more, say 55 Watts. (Actually I have not found reliable data for GPU wattage in the XPS models. A notebookcheck reveiew stated one of the nVidia cards was 65 Watts in the 95x0.)

- There are 3 mosfets + 3 chokes for the CPU. Same for the GPU. So presumably the GPU VRM components could be working hotter, but since the mosfets are significanly different for the CPU & GPU, not sure.

- A quick & dirty way to check on a relative basis is to have a power meter connected to the wall and see how overall power consumption is with benchmarks using nVidia then run them without nVidia. I don't think it will be too accurate but will give you some idea. I think some of the notebookcheck reviews provided some power consumption at the wall data

 

From the mosfet, more like 4-5mm. You can stack thinner pads. Fluffy is a good idea, but I still haven't found such. Arctic 6W/mK aren't. Maybe some which don't have a W/mK label?

I guess it is likely not enough to pad locally to the bottom case, because the whole VRM area looks overheated in the thermal images. The heat should be transferred to the cooler areas, towards the vents? Perhaps a few cm wide copper band on the bottom case above the VRM area ending just above the heat pipe ends (but with thin insulator spacers towards the heatpipes) would help? No space for another heatpipe, hehe, and lots of thick quality pads tend to get expensive.

Were the pads touching the heatpipes in these experiments (drawing additional heat to the case bottom)? I find it odd that just adding pads to the chokes and VRMs would cause more heating to the top as described, because a lot of heat must be generated by the voltage regulators also without the pads, and didn't notice airflow taking it away.

 

Problem is that 5mm of thermal pads won't conduct much heat. Plus stacked pads reduce efficiency further. From a practical point of view, those mosfets are so small and far from the case bottom it seemed tough to apply with flimsy pads without adhesive and measure proper depth...

It is odd. I think everyone in the thread was heatsinking the chokes to the case bottoms but you can do a search to confirm. I used 17W/mk pads on the 6 VRM chokes and the keyboard got really hot really fast. The case bottom was very hot.

I think what was happening is the VRMs could run harder because there was a place to dissappate heat. But note if you heat the case bottom a lot, there is a point where it will no longer be able to dissapate more heat fast enough. So without an external laptop cooler (which is never too efficient) the VRMs return to throttling.

Also the hot case caused the fans to suck in hot air. This made the CPU & GPU cooling less effective.

Real balancing game here

 

It would also be possible to install 4mm copper to bridge the gap. I thought good pads would conduct enough, with some soft poor ones for the final soft layer and paste to make it stick in place?

It would make sense things got hotter at some places if nominal frequencies at full load were maintained for a longer time (local temps near the relevant sensors reached the threshold later).

Another experiment: how about a long external heatsink, like 2.5x2.5x35cm, for the VRM area along the rear rubber foot, lifting the rear of the laptop, with the pads? Would the top still get hot?

 

But I recall many were also padding the heatpipes, I think this is counterproductive in these conditions.

 

I thought about that. I think you would need to use epoxy the copper slugs to the mosfets. Or risk they loosen and short out the system. Also, I think you would need thinner copper because the case bottom has some flex and you don't want to knock the delicate mosfets off the board. Needless to say that mod might void the warranty and is not really reversable.

Alternatively, if you are precise enough, you could epoxy some copper or alu slugs to the inside of the case bottom then brige from the mosfets with thick foamy thermal pads.

Regardless, the case bottom is not a great heatsink. But, this would get directly to what I suspsct is the issue (hot mosfets). As a general matter the chokes should be cooler & less sensitive to heat but you would need to look at the spec sheets (see that handwritten jpg referenced above for clues on part numbers). People cooled the chokes as a proxy as it was much easier to do...

Interesting idea. Passive or add a USB fan? Getting it efficient and small would be a challenge but worth a shot.

One idea is with small battery, remove the 2.5" drive. Install a fan there. Power via SATA cable. Cut a 20mm hole in case bottom for intake. Then snake flat plastic, semi-rigid non-conductive tubing from fan exhaust port to edge of VRM. Then remove that foamy pad at the back of the laptop which separates VRM from back exhaust port.
- Not sure if there is space to snake tubing efficiently
- Not sure if I want to lose 2.5" drive
- Not sure if I can tap SATA power

 

I thought simple passive, an alu heatsink instead of the book I was using to raise the rear Or a metal bar just for a test, or a fan pointed to the VRM area from below.

I don't think there is any space inside, the gaps must be very thin at some places. The vents are relatively near, but don't seem to suck any air through the narrow rear grille and over the VRM area.

 

Passive won't work as the heat will eventually find it's way round, you would need something to dissipate the heat. Best bet would be to hack the PWM feed (not the power though) and fit a small blower. One of the laptop fans would be ideal.
Anyone who would consider modding this laptop would be better off buying a bigger laptop with better cooling!

 

I wonder what would happen if you add a small strip of copper above the thermal pads that are on the VRM...have the copper reach over the heatpipes from the CPU and GPU cooler and another thermal pad back on to the heatpipes...? Then have like a small strip of soft plastic/foam or something blocking the heat from conducting on the case... then that way it would pretty inefficiently travel to the CPU and GPU heatpipes...I bet its possible..just might take a lot of work, cutting, measuring...and getting it just right.

edit... I was actually just looking at pictures of the cooler area...well actually I could still keep with the original idea of cooling the VRM via passive with the frame...HOWEVER... I'll also need to connect the frame to the heatpipes via thermal pads also... then that way the heat goes from VRM > thermal pad > Frame > thermal pad > heatpipes

Overall it might actually raise the CPU or GPU temperatures but effectively it would create a passage way for the VRM to be eventually cooled by the heatpipes which would carry away the heat but obviously the case in turn would still heat up in that area... I might have a working theory now...

How flexible is the bottom of the case once screwed down? Can you depress it by the cooler? I'm wondering since if the thermal pads are pressing up against the frame and the CPU/GPU cooler I wouldn't want there to be flex in the frame so it could potentially put additional pressure on the cooler and die's if you handle that area of the computer...

 

I think if such a strip mates with the heatpipes it is likely to bring more heat to the VRM area from CPU/GPU than take it away.
But the bottom case just over the heatpipe looks cool in the thermal image, is cooled by the vent output (which doesn't get hot before the fins). If the strip was touching the case but with a thin gap towards the heatpipes it might be able to take some heat away.

 

Here we go again

The heatsink only just does the job cooling the CPU/GPU, everything else has been covered already.

 

Sounds like if I plan to game on this...laptop cooling pad is going to be the way to go...

we need a 9560 temperature/observations undervolt repaste thread here soon....

 

But it will not help enough if you don't get the heat to the case bottom. And it is not sure if then the keyboard will run hot?
Maybe Conductonaut is actually less hassle?
(though the VRM area might still overheat?)

 

I expect with careful positioning VRM cooling with a active cooling pad would stop the case heating beyond reasonable.

 

Brain Anderson did something like that with a 1$ USB fan aimed at the f7 key. I thought he wrote some superuseful VRM cooling posts...

 

Yea. I'm liking this idea more and more... So basically my theory now is to have a 4mm thermal pad extend over all the VRM and make contact between them and the case. That way it'll dump a good deal of heat to the case... then for gaming/benchmark purposes... I'll have a decent cooling pad blowing right up in that area to supply cold air...problem solved. Should solve any throttling issues...

 

Hey guys so I went ahead and thermal padded my VRM to the case, with very good results.
Here are my pre and post padding results. Instead of a benchmark, I just had a game open in windowed mode.

Pre-Thermal padding temps
(with no laptop cooler - much better VRM temps without it - but slighty higher GPU temps)


Max VRM Temp: 87
Max GPU Temp: 78 (thermal limit according to afterburner and furmark, and very apparent just from watching temps)
Total Time: ~20 minutes


Post-Thermal padding temps:
(With laptop cooler - need a way to cool the bottom)

http://imageshack.com/a/img924/1782/r8Az8A.png

Note: Ambient Temp graph is the VRM temp, it is the first ambient temperature if you're using HWInfo, and it is the "DIMM" temp using Aida64.

Max VRM Temp: 66
Max GPU Temp: 70
Total Time: ~30 minutes

Well I didnt expect this to make such an impact on my temperatures, and my laptop cooler seemed to keep the case bottom temp in check. I was planning on experimenting with conducting the case to my aluminum laptop cooler via external thermal pads but it seems as though I wont need to (I may still just for the data).

One thing interesting is the power limit drop at around the 20-25 minute mark, not exactly sure why that occured =). The only downside is that the fans are a little more active at idle (~3k RPM), but that's to be expected. But anyways, at least for now i'm happy with the results!

 

Thanks for sharing. Could you take a picture of your thermal pads on the vrm location?

 

Could you document exactly what you used for thermal pads and what you were cooling?

And you might install "Dell Command-power manager" and use the Ultra Performance setting if you want more aggressive fans & performance.

Also, if you can get max temperatures so low, you might be able to ramp up performance a bit with Intel XTU by boosting turbo watt and timers. I did that with a 9550 and you may have the same options on the 9560. Just keep a close check on temps

 

So I wasn't sure exactly how I was going to go about it and I bought a few different sizes of thermal pads and unfortunately they have different W/mK. That being said I should probably redo so all the same thermal conductivity.

For anyone who is an expert at applying thermal pads... I apologize for what you're about to see

XPS 9560

http://imageshack.com/a/img923/6628/baU0Rv.jpg No pads
http://imagizer.imageshack.us/a/img922/5467/caVUfn.jpg - First layer
http://imageshack.com/a/img923/8168/NZj0ZS.jpg - First layer + transparent remaining layers (did not take pictures during during the process - oops)
http://imagizer.imageshack.us/a/img923/6814/l4z1y4.jpg - Final

Pads I used were:

10mmX10mmX1.0mm 1.2W/mK (probably should have looked at the conductivity rating before I purchased)
https://www.amazon.com/gp/product/B00EQ23NCK/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1

145mmX145mmX1.0mm 6W/mK
https://www.amazon.com/gp/product/B00UYTTXSM/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1

 

Hmm it wasn't letting me edit my post, anyways:

Laptop cooler: https://www.amazon.com/gp/product/B0012WXFO8/ref=oh_aui_detailpage_o08_s00?ie=UTF8&psc=1

Pressing, I do currently have Dell Command Power Manager installed with Ultra Performance, and I currently have set via XTU:

-100mv on CPU
-100mv Cache
-100mv Integrated GPU
Turbo boost power max from 55--> 65
Turbo boost power time to 96 seconds.

I haven't changed anything regarding "Turbo Boost Short Power Max" though

 

Great write up!!!! Thanks! I'm planning on doing something very similar. What is the total thickness of the thermal pads for the VRM??? Did you repaste the CPU/GPU yet?

 

Thanks Ginglymus for documenting your thermal efforts here with clear photos and narrative.

Interesting approach of stacking progressively larger pads as it reduces the requirement of precision. I thought only about stacking several tiny pads above the mosfets which would have been difficult without some type of adhesive (advantage of lower cost and maybe smaller components get better air cooling, which is somewhat mitigated somewhat by fact there is not much air cooling going around that area).

Most importanty, thermals have improved for your usage. And you have documented some thermal alternatives for others to consider

If you have time to run a few benchmarks sometime that would be helpful for others can compare (e.g. Prime95 small ffts & ROG RealBench Stress Test.) Before and after might be most useful (or not) but after results are still helpful for others..

 

The 1.2W/mK pads are probably okay, if you don't want to get the case bottom too hot.
Are they fluffier (easier to compress) than the Arctic 6W/mK - got some of the latter but aren't soft enough for the top layer so that I could expect them to squeeze if the stack is too high.

Next time I open I will try to use some pads and copper to lead the heat towards the cooler areas of the bottom case. But also try to leave some space for air, maybe remove that rear grille if it goes away easily. And maybe the internal airflow could be increased by increasing the narrow air space below the bottom cover by half a mm so that the vents would suck a little bit more internal air - washers below the two central screws, a few spacer pads on the vents, mind the sealing of the exhaust. This is assuming that there is a few K headroom of CPU and GPU temperatures, because more cooling with internal air would likely make them increase a bit. But I haven't got time for this anytime soon.

It is better to run a stress test than a game, so that the load is comparable with others. Unigine Heaven is such free example and much game-like, mainly loads the GPU. For the CPU, Prime95 (small FFT is a bit more demanding than default blend). Finally you can do both together. I don't know ROG RealBench, it would be good to use sth which loads the GPU uniformly, without disturbances due to scene changing in Heaven.

It is good to record all the temperatures you see in HWinfo64, particularly those on Dell EC, so that you can later find what triggers throttling and how hot everything gets; PL1, CPU and GPU clocks (see how high they are, if downthrottled); fan speeds.

 

I've ordered a bunch of different thermal pads just a few days ago. Should have those in the next 2-3 weeks. Going to order my xps 15 once they put it on sale for $150 off...lol come on presidents day!!! Will definitely post a cooling guide!


Sent from my iPhone using Tapatalk

 

regarding the undervolt, I need to set the value in CPU CACHE and GPU or I can set GPU in default?

 

You don't have to change the GPU undervolt, just make sure the core and the cpu cache voltages are the same.

 

Thought I'd like to share a few items I've found...
http://imgur.com/a/6EjCB
Interesting guide showing the increase in temps...

I'm also hoping to create a putty or dough to measure the exact thickness needed for the VRAM thermal pads and thermal pads for the VRM... With dell sales right now I'm hoping I'll be picking up my laptop on sale either this week or Presidents day... Thermal pads are already on their way and should be here in 2-3 weeks...