XPS 15 9560 owners thread.

Holy **** it worked.... Clocks, multipliers, timings. Everything...

 

Yes on Advanced Tab my choices say 2-channel as well, but I noticed that for stereo audio devices like speakers and headphones 2-channel is not mentioned for their format options. I suspect the microphone captures audio in mono and Windows creates a stereo signal from it. And when I test the left microphone I do hear audio equally in both ears.

But my concern is that the right microphone is not picking anything up. If you set your microphone volume and boost to 100% and slide your finger across the left and right mic openings do you get a full volume spike for both of them? I only get a full volume spike when passing over the left mic. I had the same issue on my E5550, so perhaps this behavior is intentional and each microphone is not meant to have the same sensitivity. Looking at the physical microphone array inside the 9560 the microphones appear to be of equal size and have the same number of wired connections.

 

I was watching the spikes as I ran my finger over and it did seem to be both the same. Have you updated the drivers again?

Sent from my SM-G920F using Tapatalk

 

Interesting. So I tried using the latest two versions of Dell's Realtek drivers (January 2017 and March 2017) and neither had any impact, but using the latest drivers from Realtek's site did fix the problem. It goes to show when all else fails go to the source and use the manufacturer's latest drivers! Thanks!

 

Good stuff

Sent from my SM-G920F using Tapatalk

 

It looks like Overwatch overheats XPS15 (from other people's reports). Can you guys play other games on it? Or GPU is useless without better cooling system?
I have not had a chance to install any games on mine. Anyway, I prefer desktop for gaming.

 

It can play most games but you should consider repasting the CPU & GPU prior to this to help prevent throttling, undervolting also helps. Most people get at least 100mv.
Also game at 1080p not 4k.

 

I need to check in EFIshell again as half the options were already enabled and I enabled the rest but the multipliers still greyed out.

 

Thanks. Good to know. I repasted (although original paste looked fine), added some VRM cooling and undervolted by 140mV.
It just all this high load testing with benchmarks, etc got me thinking about returning the XPS, getting macbook pro and installing Windows on it

 

Hey guys, I'm receiving my 9560 soon. I've been reading this thread and the thermals one (and many other threads on reddit etc.) and I've compiled a to do list of things:

- Repaste the cpu/gpu, keep the stock thermal pads
- Clean W10 install
- Update all drivers
- Update the BIOS
- Undervolt through TS
- Enable Speedshift
- Pray to the thermal gods that @iunlock or someone else finds an easier/garranty friendlier fix to the VRM temp issues than the copper sheet one

Do you agree with this? Is there anything you would add?

There doesn't seem to be a one size fit all guide to a good start on this laptop, sadly.

@mikelev doesn't the MBP throttle as well? Or Apple did some magic and manage to cram similar specs in a thinner chassis without any compromise?

 

Oopsy, double post

 

I am not sure
I actually thought there is a MBP version with only built-in Intel graphics. Apparently, MBP uses at least AMD 555. I think it is a weaker GPU than GTX 1050 and should run cooler. I was thinking about MBP under (wrong) assumption that XPS cannot run any GPU-intensive games due to throttling and therefore its GPU is useless

 

Last guy on here ago got a mbp is back here with another XPS15

Sent from my SM-G920F using Tapatalk

 

You're right, no one solution for all but you've done you're homework and it looks good.
Apart from the God part!


Sent from my SM-G920F using Tapatalk

 

You can get a copper sheet from ebay or such, if you don't like cutting the original one. Most other parts of that mod are reversible. Except Conductonaut - but it is not sure if that is required.

Btw lately I've spotted 1mm heatpipes
http://www.ebay.co.uk/itm/282530009805

 

I just created "air channel" similarly to the way described http://forum.notebookreview.com/thr...ures-benchmarks-xps-15-9560-kaby-lake.802345/
and so far results are impressive!
I did not cut or modify the copper shield. It is still attached to the bottom. Just separated the VRM area and fans with strips of thermal pad material.
My "air channel" is also a little bigger. It includes more space in front of the fans.Thermal pads are about 3mm thick (2 layers of 1.5mm) .
Not sure if this mod actually directs more air to VRM as intended or simply creates a bigger thermal path between the bottom cover and CPU heatsink or both...

Before the mod, running 3DMark Firestrike combined test (GPU + CPU load) was causing 80+ temps and VRM throttling after 5min.
After the mod I can run that test with exactly the same settings for at least 20min and VRM temp stays at about 76 and GPU/CPU at 72 degrees.
Adding 1 thread of FFT in Prime95 increases VRM temp to 79, GPU to 76, but still no throttling of any kind!

Testing was done with turbo disabled (i.e. CPU at 2.5GHz, 0.7v), enabled speedshift, undervolting, and heatsinks on VRM FETs. Enabling turbo raises the temps by another 2 degrees and causes GPU thermal throttling.

 

Tnx for letting us know.

Can you post some photos? Did you do any sealing from the VRM area towards RAM etc?
And run 30min Prime95 on all cores and Heaven at once?

> Not sure if this mod actually directs more air to VRM as intended or simply creates a bigger thermal path between the bottom cover and CPU heatsink or both...

I guess the bottom cover doesn't add so much cooling surface. Does it get hot at that place on the outside (old thermal images show that as a cool area, likely because of the nearby fan exhaust)?

In the original state there already is half of iunlock's dose of the copper shield in that place - he doubled it, plus made a bit of a gap towards the bottom cover (as leading much heat to the central bottom cover was shown to belong-term counterproductive).

 

Well damn, does anyone know how to reset CMOS/NVRAM without taking it apart and pulling the battery/CMOS battery. I was fooling around with the EUFI variables and somehow managed to disable SST and can't get it back, not to mention I have no idea which variables are enabled and disabled anymore. I think I'll make a table of things that actually work and what they actually do.

I already tried setting defaults in BIOS, and doing the hard reset hold power button thing. No dice.

 

SST isn't there, you usually have to use Throttlestop to enable it. Have you tried doing that?

 

I enabled it in the firmware, but I think I screwed something up or enabled a setting that interferes with it. I can enable it in TS, but it's a pain in the but to have to run a program every time I restart.

 

I have mine set to run using task scheduler. It starts up and runs minimized so I don't ever see it, but I do use it for more than just SST. I don't mind it though, I also didn't know you could enable SST via firmware. Hope you can get it working again!

Sent from my ZTE A2017U using Tapatalk

 

Can you get it to basically auto close doing it your method? That's exactly what I'd do, but I could never figure out how to get it to load up with my settings over powershell or CMD startup.

 

It starts minimized so I only see it if I show hidden icons on the Taskbar

 

Mine reset using one of the BIOS resets, I know it has 2, did you try both and I think the other one is holding the power for more than 25 seconds until it start to "do stuff" with a flashing charger light?

Do you have the remotest idea what you may have fekked with so one of us can check against our own laptops?

 

It cannot run Prime95 4 threads (i5 CPU) and Heaven simultaneously
Temps get to 78, GPU starts temperature throttling (GPU clock drops to 1000), temps drop, throttling stops, then heating starts again, temps raise to 78, etc.
Heaven is configured to run in FHD resolution, no AA, medium quality. CPU turbo disabled.
I am guessing this test might pass if Heaven is configured for lower resolution, but not in FHD.

No pictures at the moment. Will take them later.

 

I did some more testing with different Heaven settings and different number of FFT threads in Prime95. Basically if I load GPU with some benchmark (almost any graphics settings) and use more than 50-70% of CPU continuously, GPU/CPU temp will get to 78 quickly and cause thermal throttling.
VRM temp varies depending on benchmark config and VRM temps got better with FET heatsinks and "air channel", but even if VRM stays below 75, GPU will throttle down at 78 degrees anyway.
Heaven by itself creates a 50% CPU load. Adding 2 threads in Prime results in 95-100% CPU load and that together with GPU is enough to overheat everything - GPU and VRM.

 

Then, Conductonaut? :evil:
Did the GPU temperature go up after the "air channel" mutilation?
Did you use tape to seal the outer heatpipe end to the case edge?

 

Why the hell would you run more than one benchmark unless you are looking for a 15" doorstop?
It's a laptop, not even a gaming one, it isn't particularly advisable to run one benchmark for any length of time unless you are the sort of person to try and drive down the motorway in 1st gear at redline telling the recovery man you have no idea why the bonnet has piston shaped holes in it!!

 

Probably to heat everything up.

Sent from my D5803 using Tapatalk

 

No, I've burned enough PCBs in my life by dropping conductive stuff on them I am not disabling DPTF either

No, at least not in the case when CPU load is under 60% + GPU benchmark.
At 100% CPU + 100% GPU it is hard to tell what exactly is going on. All temps go to 80 or 80+ pretty quickly and bottom cover heats up too much.

No, I did not bother with real sealing. Also, there are quite a few small holes between the VRM "air space" and the rest since surface is uneven inside the laptop.

As I said, the mod helps a little when CPU load is under 60% (plus GPU benchmark), but at 100% CPU there is so much more heat that I think even if I was able to cool the system by a few more degrees, it wouldn't make any difference.

I tried Fallout 4 at FHD with medium graphics quality, 4x AA.
60 FPS, both GPU and CPU are at about 50% load on average (reported by Afterburner). Temps are reaching 70 after 1 hour, but stayed at about 65 most of the time.

 

I want to know its limits I will play with this benchmark stuff for a few days and never touch it again in the next few years. I might never even play any games on it since my desktop is faster
Also, I think you have to be very unlucky to actually damage something by occasionally running these tests with automatic throttling enabled and manually watching the temps.
VRM components should be rated to at least 105 degrees, probably 125. Yes, actual components might get hotter than temp sensor area, but I still think the chance of causing real damage is small.

 

I guess those who havn't release the magic smoke will think like this, I on the other hand have released the magic smoke and also know how poor the cooling is on this laptop to know it is a really bad idea, but it isn't my laptop so fingers crossed for you

 

That should be really simple to try, it just takes two stripes of electrical insulating tape, one on each end of the heatpipe. I recall Pressing came up with this somewhere in the long 9550 thermals thread. As much as I understand, the idea is not to let air from the vents go by on the heatpipe side, but rather through the cooling fins (or some towards the VRMs with the redirection).

 

Hi everyone.
Guys give me a small advise.
I don't want to read all 200+ pages of this thread.
I wanna buy 9560 (fhd, 32G, i7, 97W , SSD) .
From Charactiristic point of view it looks fine (I'm a developer and will use it as a workstation).
But what about usability point of view? I'm a modest person.
It should simply work. But i heard that not everything is okay. Is It truth?

 

get reading ya lazy git.

Sent from my SM-G920F using Tapatalk

 

I've had mine for a few days and I like it overall.
I think most of the heating problems described here only happen when you play GPU-demanding games. Using just the CPU at full power (let's say compiling Linux kernel) should be fine without any mods.
I am doing some cooling mods just for fun, but I believe they are not necessary for software development tasks.
If you are planning to use it for gaming, it is a different story. I believe its cooling system is not good enough for running both CPU and graphics card at full load. However, you can probably say that about any laptop that packs similar processing power in chassis of similar size.
FHD display is nice. I do not need higher resolution than that on 15 inch. There is very noticeable backlight bleed at the bottom, but I am OK with it.
For me, the main problem is that PgUp/PgDn keys are combined with up/down arrow keys. It drives me crazy. I normally use PgUp/PgDn a lot. I have another laptop for work, but if I had to use XPS 15 for development, I'd use external keyboard and external 2k monitor.

I got the cheapest version of 9560 with i5 CPU, HDD instead of SSD and 8Gb RAM and smaller 56Whr battery (about $1100). Installed 1TB Samsung 960 Evo M2 SSD as main drive, 512Gb SATA SSD for second OS (Ubuntu) and 32Gb RAM (another $800 total).

BTW, judging by your name, do you speak Russian ?

 

Yep. I'm from Russia)
Frankly i even don't think about gaming and about cooling a system.
My main doubts are a battery and a system stability (blue death screen)
PS: Information about git made me sad(

 

Just so there is no confusion, brining the sensors even close to 125*C for "brief testing" is likely to damage components and will shorten expected lifespan. You can not defeat basic physics.

I suppose one can look at the spec sheets for the mosfets and chokes for a better view on limits but remember these are in optimal lab settings so are optimistic. Also the distant, rudimentary temperature sensors are significantly underestimating component temps. . .

 

And for some reason the settings had defaulted, I think I may have reset my BIOS without doing all this again s speedshift was off.

All working now

1st post updated with the link to the relevant post about the XTU EFI hacks.

 

While the options may no longer be greyed out I can't increase any CPU multiplier.

 

Haggism & GoNz0- does this mean "exposing" these factors in XTU is interesting but will not permit improved performance?

Thanks!

 

Well, you can't increase core multipliers so no point in those, it seems everything that may ask for a reboot will not work so a wild guess is the vars just enable the options but another var will be needed for them to work, or it could be locked down so nothing can change them.

As for the core voltage sticking after a reboot, well it always did that, the question is if it will stick after the so called dirty bit is set to inform XTU the system crashed so it removes the undervolt. Despite mine not crashing it still removes the undervolt within a few days so I will keep an eye on it now to see if it does reset at some stage.

 

I am not sure what exactly you are proposing to seal. Is there a thread/post about that?
What is the goal? To prevent air from going around the fans towards memory/touchpad area along left and right edges of the laptop?

 

I don't know about battery life yet.
What system stability problem are you talking about? I have not had any problems.
What information about git? I've read many posts/threads on this forum, but must have missed it. Are you talking about source control tool git ?

 

The chassis has foam strips to stop air leaking around the fans so I always find the mods where people stick tape over the fans to be pointless and if any adhesive is exposed the dust will stick to it restricting airflow over time. You will find various threads relating to the 9560 within the 1st few pages of this forum as they get replies often enough to keep them close to the top.

 

Sure, 125 cannot be a rating for continuous operation, but devices with 125 max rating should be able to operate at 85 without damage. Not even sure what the actual temp rating for VRM. I would not assume higher rating than 105 degrees (125 is more common for automotive, military grade, etc).
Yes, underestimation of the actual component temperature is totally possible and even likely, but not by 20 degrees.
Throttling starts (and heating stops) at some 80+ degrees (reported by sensor). Stock XPS15 can probably reach the same temperature (80+) by playing GPU-intensive game on a hot day (high ambient temps) without any benchmarks. Benchmarks just get it to 80+ a little faster. If that temperature damages components, we should see many laptops killed by occasional gaming.

All that said, I would never recommend stressing any laptop by excessive heat. There is definitely a chance for thermal damage. I just think it is not that big. Anyway, I wouldn't do it with somebody else's laptop

 

I'd like to use SpeedFan to control the fan speed, but this usually requires a BIOS setting to fix the fan speed. Otherwise any BIOS or OS level fan control will override my settings in SpeedFan. I've tried Dell Power Manager, but it doesn't provide an option to set a fixed speed for fans. Has anyone figured out a way to have fine-grain fan control on the 9560?

With Speedfan I can specify a linear relationship between fan speed and temperature and set the min and max for both, but these setting get overridden by the BIOS/OS level fan management.

Also, SpeedFan usually has the ability to disable the BIOS/OS fan control on its sensor settings page, but none of the available sensor settings accessible to SpeedFan for the 9560 are able to disable the BIOS/OS fan management (in my tests).

 

Nothing you can do unless you know your way around the EFI shell and what bits to flip, should you need to ask what I just wrote then I would leave it well alone.

 

I have now concluded that the latest EFI hacks are a waste of time as the voltage reset again.

 

I made a few EFI changes on my E5550 to enable device support with OS X. However this was using a combination of GUI based tools inside OS X and not bit level modifications from a command prompt.

But the fan issue is big enough for me that I would be willing to work with someone more experienced and do any number of tests to figure out how to solve this. I would only need to modify the EFI to set the fans to fixed speed (preferably to 100%) to avoid the chance of overheating if SpeedFan fails.

I'm sure many other users would find this solution helpful, since the current delta for fan speed changes is very poorly granulated (i.e. 0->50->100% instead of 0->10->20->30%...).

Also this would work well with any of the popular thermal modification ideas in this forum since they all depend greatly on the airflow provided by the fans. This modification would hopefully lower overall fan speed and make fan speed changes far less noticeable in those scenarios.