XPS 15 9550 temperature observations (undervolt + repaste)

I still haven't removed those thermal pads, no issues at all for me.

 

I'm asking because I've confused a lot in all those pages, and not everything seems logical to me. I don't think that the insulation at the center of the grille is to prevent hot exhaust from being sucked back into the machine, the hot exhaust should not be near. Due to the vents, that opening would not be exhaust, rather only increase the intake port, indeed drawing a bit of warmer inside air, but I guess not so much, the majority of air would take the shorter path through the current openings?
I recall most of the failure reports were with padded heatpipes? Or was there any just with the grille open?
I'm also not sure if this would in fact cool the top VRM area much, as it is not in the straight path from the grille centre to the vent intake.

How about the small plate near the right vent, is there any image what's below (I recall someone wrote another VRM area), and what is supposed to be the function of the plate, heat sink?

The low-grade thermal pads sound reasonable to me. I don't mind if the bottom is a bit hot, because I surely wouldn't use a laptop with vent draw from below on my lap or on some textile without a hard surface below, would need to clean vents all the time. But I wouldn't want to get the intake air much hotter, as much that it'd reduce cooling efficiency.

 

See posts above by GoNz0, professional laptop engineer for major PC makers with over a decade of experience. Feel free to ping him with more questions as he is most qualified to provide you more details based on experience.

Better yet, try removing the foam and stress testing with benchmarks and games and post the before-after results. That would be interesting and end any speculation.

The hot exhaust is expelled from that same grille, just to the left and right of that center grille insulation.

 

I intend to try out whatever looks best. But I'm not sure I fully understand what looks best, don't have much time to play, and don't want to repeat others' mistakes I may have overlooked so far.

I understand the grille on the bottom cover is the intake, while the exhaust is hidden in the gap between the top of the keyboard and the display hinge, next to the grille at the ends of the heatpipes.

 

Think about it, that air that's getting forced out the vents hits a brick (glass) wall and gets forced out the way, I assume they found out if also got shoved back in or they wouldn't have put that cover in place to stop it.

 

Are we talking about the same grille?

I guess you mean the one between the vent exhausts below the display. On most laptop the exhaust is out in the open on the edge, with no intake nearby. Here it is obstructed by the display, and the central part could actually act as an intake. It is odd that there is an opening at all, I'll check if it is blocked decently when I open, can't seem to find good pics. Or maybe a small opening in the centre would be okay?

But I was referring to the grille of the bottom aluminium cover, which is also open only above fan intakes but blocked in the central part. On a flat hard surface, exhaust shouldn't get near, because of the back rubber foot. Though the incoming air might get heated if we raise the alu cover temperature via thermal pads to the VRM or even the heatpipes. Btw the heatpipes run just there, so pads on top of them would close the grille too.

Found the image with the area next to the right vent stripped and padded
https://www.reddit.com/r/Dell/comments/4zajkk/dell_xps_9550_thermal_padding_mod_update/
https://i.imgur.com/ED8MmpJ.jpg
Says throttling without a cooling pad under the laptop though.

 

I think the bottom full length grill is more for aesthetics the same as the 9530 as that had the same design and internal cover. it will no doubt be there to help direct air in where the fans are.

 

But why do you think they blocked the bottom the centre?

 

I thought I gave my answer? It is a full length grill to look pretty.

 

I just saw Iin another NBR thread that the 9350 had the ability to configure TDP hi/med/low.

Looks like that was via the power options>advanced settings>Intel Dynamic Platform and Thermal Framework Settings

Did the 9550 ever have that option on with older versions of Intel Dynamic Platform and Thermal Framework Driver? Could it also be a hidden option in the registry like we see below?

https://www.reddit.com/r/Dell/comments/534uah/new_9350_drivers_thermal_framework_1820a_wifi_and/

Offender in the 9250 FYI :
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Power\PowerSettings\48df9d60-4f68-11dc-8314-0800200c9a66\07029cd8-4664-4698-95d8-43b2e9666596

 

TDP has one option 45w. Waste of a drop down menu tbh.

 

Does anyone know how to keep throttlestop from resetting after putting the laptop to sleep? I noticed that the undervolt seems to reset after sleeping, where as restarting doesn't.

 

Do you have it turned on? I've never experienced that.

 

Yes I have it turned on. Does throttleStop have to be kept open all the time?

 

Yes, of course. Leave it on in the tray.

 

Thanks for the reply

 

someone tried to use the Liquid Metal?

 

Yes - search this thread...

 

Today, I did some stress testing on my 9550 (6300HQ) for fun and it shocked me as the laptop did not spin into PL throttling with ROG RealBench StressTest. That always killed my laptop even with max thermals below 75*C. I am trying to figure out what changed.

1. Best guess - Last month I was messing around with the newest version of Intel XTU software which lets you bump up:

- turbo boost power max
- turbo boost short power max
- turbo boost power time window

==> Bumped "turbo boost power max" to 65 watts (from 55 watts). That 65 watts shows up in HWiNFO64 also.
==> Also increased the "turbo boost power time window" to 96 seconds.
==> These settings seem persistent as they stick in registers even after several weeks of shutdowns.

Notes

- several versions of Intel XTU have CAUSED power limit throttling so after setting above factors, I don't have that buggy program running on my computer. Set it and never open it again
- XTU also will conflict with ThrottleStop so those two programs can not be run concurrently
- I tried adjusting those items in ThrottleStop (turbo power limit window) but don't think the settings did anything on my system

** DON'T EVEN CONSIDER INCREASING TURBO SETTINGS IF YOU HAVE NOT UPGRADED CPU & GPU THERMAL PASTE AND CHECKED FOR CORRECT CONTACT OF VRAM THERMAL PADS

** TRY AT YOUR OWN RISK. SEE DISCLAIMERS IN XTU.


2. Other changes in laptop that may have improved performance

- enabled "Ultra Performance" (fans-processor boosts) via Dell Command-Power Manager
- enabled SpeedShift EPP=0 (via ThrottleStop)
- disabled c1e states (via ThrottleStop)
- updated BIOS, drivers, Windows 10

fyi I always run ThrottleStop with undervolt to CPU and GPU
X-post because this info may get lost in original esoteric original Power Limit Throttling thread

 

Maybe a new bios fixed the PL throttling issues? I haven't tried to stress out mine lately.

 

Not sure Custom90GT. I also haven't done any stress tests since the summer so also think there is a high probability Dell did some BIOS work under the hood.

I am also skeptical that all the protection gates from Dell and Intel would allow a 55watt system be reconfigured to 65watts for a long time. But that seems to be the case. But I am having a tough time getting this to throttle. Obviously power reconfiguration has the potential to kill any laptop, especially those without user thermal upgrades highlighted in this excellent thread...

What is certain, is that I am having trouble forcing the laptop to throttle without resorting to very high temps or dangers of FurMark etc...

 

today arrived my new xps 15 9560, I think i'll open it tomorrow

 

Why tomorrow? Wait a week or so.

 

just to see if it is equal to 9550. Tomorrow I get the tools from amazon. for repaste I wait a few day.

 

Sorry I was being sarcastic. I can tell that English is probably not your first language. Let us know what your temps are like.

It seems like everyone who ordered a 9560 wants to wait a couple of days before opening it though, I can't figure it out.

 

sorry, it isn't my first language . In idle status the temperatures hover Around 41 degrees on i7. I have not tried to do a stress test, I'll let you know.

 

Hah don't be sorry, it's great you speak multiple languages, sadly many Americans can't say that. I would be curious to see what stock temps and speeds are while running prime 95. My guess is they will be similar to the 9550 and then greatly improved with a repaste/undervolt.

 

with OCCT and prime95, the temperatures hover Around 80 degrees.

 

That's actually not bad. Did you happen to notice if the CPU was throttling and decreasing clock speeds?

 

I did not notice any throttling and decreasing.

 

Maybe they're hoping they wouldn't have to. I mean, having to do this is a bit ridiculous, it should've been done properly at the factory. Better paste and pads shouldn't cost them more than $2 per piece in volume production, and they should be able to pick them if you are.
Maybe everybody's busy reading through 68 pages of this thread etc
I've waited a couple of months before repasting and repadding mine, mainly because there was always some stupid issue that made me consider returning ;(

 

I assumed he meant open the box and take the laptop out... I've seen several posts where the laptop arrived but they weren't going to even try it out for a couple of days.

 

It could be they are excited to have one but also know they have a 75% chance in something being wrong out the box after reading the nightmares on here and want to delay the heart sinking feeling that dead pixel brings

 

So, I've been trying to figure out if opening the rear central grille between the fan exhausts, in the gap below the display, helps with "VRM" throttling. From the factory there is a tape over it on the inside, but not covering it completely.

I've repasted, repadded and undervolted (i7-6700HQ: -150, -150, -100) recently. It seems to be done okay. It can run Unigine Heaven without throttling. Then I added Prime95 (blend), and after a while throttling of CPU did show up, although CPU and GPU temperatures were not at the limit. GPU fequency was max 1188, and the CPU was being throttled between 1.5GHz and almost up to 3.1GHz.
So, I was trying to block the un-taped grille from outside and then reopen it, and see how it matters. The conclusion is that I can't report something meaningful. The throttling is not stable, there is a lot of oscillation. Maybe due to scenes changing in Heaven... It doesn't seem to make a huge difference though. Lifting to rear by a couple cm seems to help towards higher CPU frequencies.

 

That looks like power limit throttling. But some people told me it is not. I found two ways to address that in my i5:

A. VRM seems to be an area of stagnant air that continues to heat up with no place to go. Obvious solution is to

1) create a cool air intake port,
2) insert fan (in a desktop convection via hot air rising could work but I don't think that will be effective in a small laptop)
3) create hot air exhaust port.

Problem with just removing that covering at the bottom case between the fan intake grills that you only solve 1 of the 3 steps.

Another solution is thermal pads between VRM chokes and case bottom. That can cause some real issues noted in this thread (keyboard get super hot and intake air for CPU/GPU gets super heated and thermals spiral out of control). But one or two guys seem to got it just right with cheap thermal pads.

B. Also I noted the other day, you can use Intel XTC to increase Turbo limits (watts and time) on the 9550. I can't get my laptop to PL throttle using that trick. You need to use caution by watching thermals don't go out of control. I was really happy with this. Not sure it will work on a 9560. See my prelim comments above. Also there is a power limit throttling thread here for 9550.

 

I surely agree the essence of the problem is with the buildup of heat which doesn't seem to go anywhere much.

Then there is obviously a feedback loop to downthrottle the frequencies, which is kinda slow and oscillating (mind it may not oscillate is all conditions). It is surely meant to protect the laptop, so I guess shouldn't be tampered with, we can only try to avoid it having to kick in. It would be easier to analyze stuff if we knew which sensors are used and how exactly, it likely isn't rocket science. We shouldn't assume it is done perfectly though. I mean, when the frequencies go higher, some of the temperatures, which probably aren't involved in this feedback, go higher than what they were before the repaste, and this may not be good.

I didn't do the grille at the bottom, because I thought air wouldn't move past the VRM area. I lifted the tape partially covering the grille at the rear in the gap between the laptop body and the display, between fan exhausts. Mind this should be an intake not exhaust, because the vent exhausts and heatpipe fins are sealed, sort-of (I used some rubber foam to close the gaps better).
I didn't mention that I recycled the original pads from the VRAMs and the SSD stacked in layers to pad the VRM chokes (R22) to the case bottom. What was left I put on a shiny chip near the right vent and the battery, and two other chokes. One failure mode of padding could be that if padding is too thick, the sealing of the fan exhaust by the case bottom may get less tight. Again fluffier pads would be better for this.

 

I solved my throttling by defaulting settings and nuking XTU. XTU was causing it, and by switching to ThrottleStop and undervolting with that, it was fixed. I also put 1.5mm pads in between the back of the case and the top of the heat sink.

 

this is 9560, I've noticed that on the heat sink on the screw on the left there is a black sticker, probably for the warranty.

 

That's interesting. Guess Dell doesn't like that we fix what they can't. Anyone else with the black sticker?
Is it possible to remove it and re-apply it?


#XPSstickergate

 

That looks for all intent like adhesive felt, probably to stop it rubbing or making a noise if the bottom of the case is pressed in.
If it doesn't say warranty void or was never fitted *cough* I wouldn't worry, I doubt the 3rd party engineers would blink twice removing it.

 

Hey guys, I did my best to read the majority of the forum.. made it to about page 40 at 5 am last night and need to start studying soon. Anyways, I recently bought the XPS 9560 and thought i was experiencing strictly GPU power throttling, as my CPU and GPU temps are relatively low (TurboBoost currently disabled via BIOS). However, since reading about the DIMM sensor (or rather, the temperature sensor that GoNz0 had pointed) I downloaded Aida64 and lo and behold, I believe the throttling is coming from that sensor as many others are experiencing with the 9550. My CPU temps and GPU temps seem to be just fine, however once that DIMM sensor reaches 100 degrees C it causing my CPU and GPU to throttle (GPU is dropping from ~1700 MHZ on MSI Afterburner to fluctuating between 800-1200 for quite some time).

So, my question is what is the current update on this? I am a little hesitant to void the warranty since I just bought it last week, however it seems that people are placing thermal pads ? Not really sure if I should return it, or what. If this happens to all of the XPS 9550's and 9560's than it seems like there would be no point in returning it.


Aida64 DIMM Temp: http://imageshack.com/a/img923/6868/9kixQi.png
Edit: Each peak is when the throttling occurs...but i'm sure thats fairly obvious =)

 

Whilst electric components vary, the 100*C range is a real problem for many laptop electronics. Performance declines and you risk shortening component life a lot (if not killing components outright). Try not to hit those thresholds.

1. Undervolting the CPU might also help reduce heat inside the laptop. It typically dropped CPU temps by a staggering 10*C+ in the 9550. That likely will be a bit less effective for Kaby Lake processors than it was for Skylake but worth a try.

2. At least one poster said the 9560 thermal paste was a mess. So replacing the thermal paste on your GPU & CPU might help by allowing the fans to extract more heat from inside.

3. The VRM area of the 9560 is different but seems to have the same issues faced by the 9550. Not sure if anyone has tackled that yet.

Regardless, if I were in your shoes, I would try #1 and test. Then try #2 and test (always carefully monitoring all temp sensors).

Most of the bugs are worked out of the 9560 so a bit of thermal work and you can get a good boost in performance for 10$ and a few hours of work

 

Thanks for the reply, is there any warranty stickers I need remove to access the GPU and CPU? Also, the throttling occurs relatively quickly. 1 Benchmark in Unigine Valley it throttles, and even only a few minutes of just 1 Overwatch match. I typically start seeing the FPS drop ~20 in the first 5 minutes - this was confirmed while viewing MSI afterburner log afterwards.

Edit: I also want to mention im using a zalman laptop cooler

 

Don't know. There were no warranty stickers on the 9550. The photo of the 9560 shows black tape over one heatsink screw; that tape does not look like a warranty item to me. One of the people here said it might be to prevent rubbing noise between the heatsink and the case bottom.

Component temps can hit 100*C in just a few minutes throttling could simply be a thermal issue. It is tough to tell because I think PL Throttling is not only based on a simple watt factor but has some temp factors (partially related is fact VRM performance drops off a cliff as things get hot).

If I were in your shoes, I would get thermals optimized. Then if there is still PL throttling but your laptop is running cool, look at CAREFULLY boosting turbo watt & time limits I noted above a few days ago...

 

Aida64 reads it wrong it is the VRM and DIMM sensor he mixed up.

The laptop knows what the sensor is.

What you could is post some dumps on the AIDA64 thread I posted in to get this fixed.

https://forums.aida64.com/topic/3344-dell-xps-9550-wrong-sensor-info/#comment-15367

To create the dumps aida 64 needs the status bar active (under the view menu)
Right click the bottom right of the status bar and under sensor debug do a smbus dump then a isa sensor debug.

Attach them to the thread explaining this is the refreshed 9560 and he will add it to the next beta or ask for more details.

As for the heat you will not void the warranty if you do any work, what you do is your call as we have various opinions on the matter

 

No warranty void on this, they make the service manual freely downloadable and have asked customers to open up the laptop n the past to check things.

 

Ok thanks guys, unfortunately I have a **** ton of school work to do. GoNz0 I will try to do that when I get a chance, also using HWInfo64 that DIMM temp corresponds to Ambient temp in HWInfo64 (which is the VRM sensor?). I suppose the first thing I will do is well... buy a small enough torx bit, and two repaste the CPU and GPU.


I do have to say isnt is strange that the CPU core and GPU temps are fairly low, yet the Ambient/VRM temp is super high? Highest GPU and CPU Core temp was 77 C and 86 C respectively, while the Ambient/VRM reached 101 C.

Thanks again guys, I will post my results at some point here soon.

 

For the record GoNz0 is a professional laptop engineer.

When you read the 9550 threads, GoNz0 seems to advocate a good CPU & GPU repaste. But may classify as heresy certain activites such as: undervolting, replacing non-defective VRAM thermal pads, using thermal pads on VRM chokes, boosting turbo wattage & time limits.

Like many others, I am a mere laptop enthusiast, so you should weigh our opinions accordingly.

 

"Wera 05118181001 2067 HF Kraftform Micro Screwdriver Torx Holding Function TX5"

Best for the job, never stripped one yet using this

 

I'd start with software, then proceed to hardware. If your CPU and GPU temperatures are low, there is a chance you don't need repasting that badly.

It isn't strange if you see the thermal image, the VRM area seems to be overheating as it isn't being cooled actively
http://forum.notebookreview.com/thr...0-owners-lounge.784473/page-102#post-10347023
There is a grille at the rear between the vent exhaust, where some air could be sucked in for cooling. But it is partially taped-over from factory, and also if the tape is lifted there isn't much difference, the vents don't seem to suck much air there.

It appears that the cooling system is too weak when both the CPU and the GPU are loaded heavily. For gaming, some report that the experience is better if the CPU is downthrottled to produce less heat so that the GPU can run freely. This can be done either by limiting the CPU clock in Power options or using ThrottleStop you'd already use for undervolting
http://www.ultrabookreview.com/10167-laptop-undervolting-overcloking/

Lifting the rear of the laptop by a few cm, in such way that the exhaust air from the rear can't easily get to the fan intakes at the bottom, seem to help a little bit.
I've made a few small stacks of old thermal pads from VRAM and recycled them on VRM chokes so that some of the heat gets to the case bottom like some suggested. Reportedly it isn't the best idea to use the best pads there because then the case gets too hot. Maybe an external fan there would help.