XPS 15 9570 Owners Thread

I already tried all profiles, they seem to be ignored.



Though after setting it to "max power", vcore of GPU is now a "bit" higher but still totally broken.


Why does it always say limit reason "pwr"!? is that normal? It somehow doesnt realize the GPU load is 99% and boosts up correctly.

 

what happens if you use Windows' "Balanced power plan" instead of Dell's?

 

There is no Dell plan, just balanced, because I did a clean Windows 10 installation. The Dell Power Tool also doesnt seem to create a new plan after installation.



It is using the Windows balanced plan.

 

Does removing intel DPTF improve the situation?

 

I welcome your or others help on this, really. I dont think, that is an appropriate approach though and I dont think that it would work. Even if I remove the Dell DPTF package, the drivers would remain I think, because they are included in Windows. And there was no change in the past months on the DPTF drivers. And I am, as it looks like, the only person on here, who has this issue, right? I would welcome people try maybe with latest Windows 10 updates, if their laptop works normally still in games.

Any benchmark you would want me to do other than FurMark (it should still work I am sure, but it doesnt). Whats your result on GPU in Cinebench15?

Whats your vcore for GPU in P0? Any other way to find out, if maybe something isnt working normally with power management? Sounds implausible, but maybe something on the mainboard broke and GPU power isnt working anymore. As I see it though, it is totally some stupid software (Windows) issue. Because I had a full boost about some hours ago, after I toggled countless times back and forth out of a game and then, randomly, the GPU boosted to 1709MHz... which it also shouldnt btw, because the game I tried, worked normally before with a 1150MHz clock and 60FPS.

 

The normal operating point is around 1V/1700 Mhz. I get 105fps at stock in cinebench opengl.
I suggested that since it seems a power management problem. What is the latest windows patch that you have 2018-08 or 2018-09?

 

This is totally nuts. I rebooted Windows a few times and opened WoW several times at first after login and 1 time out of 5 reboots, I got a 1709 boost and 60FPS, but the others all were just 800MHz. Waiting and closing WoW and then reopening also then resulted in the broken state and it never boosts up anymore from then, in any game.

The Windows updates of the last days are KB4100347, KB4457128, KB4456655

 

I am using 398.36 as drivers

 

I actually cant remember when I updated to 399. What's the correct way to downgrade, thats like the last straw I have right now. I hate this laptop. I am literally desperate right now.

I just noticed, that when I start the game I mentioned above in the one screenshot, which is btw "The Awesome Adventures of Captain Spirit", I get normal boost speeds during the loading screen, then when the game is fully loaded, clock breaks down to 800-900 and I have 17FPS. When I then tab out of the game, and tab into MSI AB and wait for about 10 seconds, the GPU suddenly boosts up to 1709 while having AB as focused app open, when I then tab back into the game, the boost stays with 1709 and I have 60FPS... UGH. However, just last week, I was able to play the game with 60FPS with just a GPU clock of 1100MHz and the fans were SILENT.

The alt+tab workaround doesn't work in WoW though because it has no fullscreen mode anymore.

This is just unbelievable.

In Cinebench I cant get more than 32FPS though whatever I do, WoW boost works maybe once in 10 Windows reboots. Other games I tried like PolyBridge, They Are Millions, all broken.

Do you use 398.36 from the Nvidia site or the Dell provided package? Which is though from 26 Jul 2018, a bit old.

 

Are you on the latest Bios?

I didn't have any of these issues until I switched to 1.3.1

My theory is it's that Bios because I never had those issues until then.

 

I dont think it is the bios, I am on 1.3.1 from the day it came out and thats way before since the issues started, and games were working normally last week.

Ok tried downgrade to 398.36, didnt change anything...

 

Weird, the power limit throttling makes me wonder what is going on. Perhaps reinstall the DPTF drivers? I'm on 399.24 and don't seem to have any issues. I'll have to play around a little more in a bit though, when I can get back on AC power.

 

Sometimes after a video driver crash NVIDIA restarts it with lowered limits.

SIDE NOTE: I noticed that Dell reduced the MAX temp limit from 78c to 75c.

 

Just checked mine and in Cinebench R15 I get 54.31 FPS and it boosts just fine. I also have KB4100347, KB4457128, KB4456655.

Intel drivers - 24.20.100.6286
Nvidia Drivers - 399.24

Throttle stop EPP set to 128

I even restarted 2 times and ran Cinebench and got similar scores.

 

54 fps? I get 105.

Sometimes I see odd behaviors too. Maybe they did screw things up with the latest BIOS/DTPM/Windows whatnot.

 

Ok there is totally something MESSED up... I have no idea whats going on here. I actually read in another older 9550/9560 thread ( http://forum.notebookreview.com/thr...nchmarks-xps-15-9560-kaby-lake.802345/page-28 ) that the pwr limit indicator actually hints at a VRM (overheating) issue. So maybe this is again happening... but the temps are just so low around 47°c I doubt this cant be. I also cant see how, I have Vsync on (60fps capped), AA off, render frame limit set to 2, in all games all to low. This worked before amazing and I had 60FPS and low temps, but now it's all not working anymore.

And now it gets weird: I randomly played around with Cinebench for a while, and suddenly, all boosted up to 1709 and also 4GHz and I got a 103FPS bench, then randomly it clocked down again several times to 900MHz and I got a bad one around 38FPS. The first boost actually was triggered the moment I disabled turbo boost while the bench was happening via TS, so CPU clocked down to 2.2GHz and in that moment the GPU boosted up. I activated Turboboost again and then I got a stable 103FPS bench with good temps around 67°C or so.


Temps are just fine around 60-70°c, so there is something fishy going on here... this is like a TYPICAL issue this laptop has to put on the list of countless problems.

I started WoW and suddenly, good FPS, stable 60... AND clocks were around 1100-1300... not 1709. Tested other game, suddenly worked too.

Rebooted, waited a bit, started WoW... broken again at 900MHz (99% usage, 25FPS) and wont boost up... I wonder if my VRMs are bad (maybe indicated already of the coil whine I have), and then randomly fail to give proper voltage to the GPU. Or its a whole stupid Windows bug...

And this is what I meant btw with what I said before:

Suddenly stable 60FPS in WoW... with an incredible low GPU and memory clock... and fans are silent, temps are awesome. And GPU load is suddenly also not 99% anymore, with an even lower clock speed.

 

I may be having the same issue. I thought that it was boosting fine, but it runs 90% of cinebench @ 1GHz and then boosts to 1.7GHz. My guess is that it's an nvidia bug...

Did you do a clean install when you went down to the earlier drivers @maffle?

*on edit*
running it back to back I get ~104fps as it boots the entire time. Something weird is going on.

 

No, I always just install Nvidia drivers directly and let the Nvidia driver update/remove the old one, it always worked in the past on all my PCs.

I actually did this now in the past 2 hours:

Removed 399.24 via Windows program manager (started Nvidia setup), so I didnt use DDU.
Then I installed 398.36. No improvements.
So I installed 399.24 again.
And now it is "working" again or lets say, it is better working, and I have no idea why.

It seems to be some nasty bug or VRM issue to me. Either the VRMs are overheating or are bad, or the GPU gets confused and doesnt request the required voltage always correctly, and then stays at 99% usage with a low clock. It doesnt seem to help to disable Turboboost of the CPU in that case too, and also not what is selected in the Nvidia panel. It is totally random so far for me if it works or not.

 

I don't think it's a VRM issue nor a temp problem because when it happens the system is still fairly cool.
And cinebench opengl is not a brutal test.
There are 4 throttling thresholds for the GPU, both reported by Afterburner and HWInfo.
I believe it's the GPU power limit threshold. Who sets this value and why is set so low, I don't know.

Also remember that afterburner pools the sensor every 10 sec by default.

 

You cant compare system temps with VRM temps, they can mostly get hot without having a relation with system temps. I also dont have mine "modded", so I dont have any coling straps on mine. Fact is, that the GPU doesnt seem to get the required voltage most of the time, for whatever reason. How can AB show the throttling reason? The reason is shown on the one screenshot by GPU-Z, pwr throttling mostly meaning not enough voltage.

Every 1 second, by default.

I actually enabled power limit in the OSD and it randomly is triggered all the time and when that happens FPS drop to 47 from 60 and usage goes up to 99%. Where some hours ago, it was mostly triggered all the time...

 

Some info about Win update KB4100347 posted here http://forum.notebookreview.com/threads/windows-10-redstone-4.813941/page-51#post-10795353

From the link in my post...

 

@Papusan I am totally aware what KB4100347 is, and I find it disgusting too, that it was/is distributed via Windows update, where MS said during the past months it wouldn't and would just stay optional. But I dont see any relation (so far) to the GPU boost issue. I think I will deactivate Spectre and Meltdown protection again via InSpectre.

Fun facts on this also: https://www.techpowerup.com/forums/...cking-is-lost-wtf-microsoft-and-intel.247595/

Someone in the post said deactivating Spectre protection via registry doesnt help anymore with this MC, or uninstalling them leaving behind the MC btw.

 

I don't say it's any relation between my posts vs. what I see here. But if people start seeing weird results they need to know they are in fact Micro$h4t's unpaid guinea pigs. And I have already seen the link you posted

 

@Papusan The problem with the MC is, that even if you set the Spectre Protection flag to disabled, the MC is still mostly bugged and will cause unseen consequences/effects on the system. MS actually said several times, the Mc update will be optional, and now they just pushed it anyway. I wonder why. Maybe there is some attack happening in the wild, and they had to push it.

 

We talk about Win 10 and how M$ use Win 10 users as guinea pigs... If so, then they would force this on Win 7/8.1 users as well.

 

It's probably unrelated to the reported GPU issues, but I updated my TS to the new version 8.70, and my CPU got stuck at 0.8 GHz. I tried doing several things, rebooting several times, and the CPU stayed throttled to 0.8 GHz. I downgraded the TS back to 8.60 and the CPU went back to normal.
Is there some specific TS update procedure I don't know about, or does the new TS 8.70 have some throttle issues?

 

So I did a clean install of the nvidia drivers and it boosts just fine again. Sounds like it was the timing of updates or the nvidia drivers themselves. Either way we can't blame Dell on this one...

I just tried 8.7 and it works fine for me. I did remove the .ini settings file and re-entered my undervolt values. Perhaps try that?

 

German review
i7 + UHD
https://www.notebookcheck.com/Test-Dell-XPS-15-9570-i7-UHD-GTX-1050-Max-Q-Laptop.317254.0.html

 

How, you mean you used DDU in Windows safe mode to remove it and then installed it again? Maybe I will do that too and look over the next days if clocking works normally or I fall into the issue again.

That is really interesting though, because I asked about this several times as a feature, and I was ignored so far. I would love a feature in TS actually, where you can force the CPU into 0.8GHz and keep it there. That is not possible so far in my understanding, would be a neat feature though.

 

Post this also on the ThrottleStop thread

 

Can't you just set the multiplier in the main Throttlestop window?

 

Sadly no, a forced fixed (lower) multiplier and clock mod don't work on modern Intel CPUs (up from Broadwell maybe, idk) anymore, they are ignored. I dont speak about overclocking here.

There are situations, where I want to cap the frequency though, for example watching a movie or browsing the web, where 0.8-0.9GHz would be totally fine on the 8750H. I dont want it to up clock all the time between 1-2GHz where there is no need for. Even if you set SST to 255, it does so.

I know it is possible for the OS to force the CPU into that state though, because Ive seen it be done in the past, so there must be a command for it, mostly through DPTF. TS cant do it so far though, and I would love it could.

It is really annoying, that with modern standby, you dont have an energy saving profile anymore and have no control over your laptop, as it used to be in Windows. Because I want to decide how my laptop behaves. Even if I have it on AC, maybe I want it to stay in a maxed energy saving mode possible, for several reasons.

 

You can delete the setting file for ThrottleStop when it is not launched. Then restart. Then manually set ThrotleStop again. That typically works.

 

I actually just disabled the dGPU and uninstalled Nvidia driver to see if things could improve, and so far I feel that they have although I have yet do do any "real testing". Not sure if Optimus is disabled also then since that is an nVidia technology. I have no plans to ever game on this BTW.

 

There is deal in the UK going on, £1,488.00 for an XPS 9570 (16GB RAM, 512GB SSD, FHD, 97Wh battery variant) with 2 years support + 2 years accidental cover included.

I am very tempted to proceed to purchase at this price, any naysayers?

 

Erm...

 

I guess you are missing the point here, of course there will be no GPU switching if there is only one but that does not guarantee that Optimus itself won't interfere in someway trying to.

 

No I think it is the opposite around But no bother, if you want to do that, fine. There is no problem with Optimus (jump back and forth, and if you configure it correctly), that is long fixed.

 

What's the opposite, Intel Optimus? End results are a bit weird though, the fans are ramping up much more than before with out the Nvidia drivers and the GPU being disabled indicating that only the dGPU would be used. At the same time, the Intel control panel is happy to tell me I am using the Intel iGPU. Have not yet looked if you can disable this in BIOS instead?

Actually, main reason for even testing this out are the frequent crashes together with the TB16 dock that is driving me crazy. We are talking deep level crashes here, when even the mouse cursor starts moving perhaps 20 sec behind actual movement with fingers on the TouchPad.

Sent from my SM-N960F using Tapatalk

 

@improwise I have set "use iGPU" in the Nvidia panel as default, and just add apps/games to the exclusion list to use dGPU. This way I never had any issues at all with Optimus. Just disabling it/removing the driver is not a good solution, why did you buy the laptop then anyway?

The Dell TB16 is just horrible, and the price is one of the biggest jokes ever told. I wouldnt even pay $100 for it. A fair price for this would be ~$50, not a single penny more.

 

Okay this is now getting very hard to turn down...

XPS 9570
4K Ultra HD
Intel i7-8750H
32GB Ram
1TB SDD
3 year warranty
3 year accidental cover

£1,961.51...

I have until 26th September to pull the trigger, and with only Whiskey Lake 15w laptops coming out, is it worth me sticking with my XPS 9360 another year for next years laptops, or pull the trigger on this?

 

@iMbaQ we can't make these calls for you. You should weigh all the pros and cons and make your decision.
Do you NEED it? Probably, not. Your laptop is just one year old or less. Do you really want it? If you have the money, go for it.

This is coming from someone who changes laptop only once it's incapable of fulfilling my workflow (that means 4+ years).
But I understand that there are people that who want the new thing and have the money for it.

 

I have the base model of the 9360 (256GB, 8GB RAM) so the performance difference will be immense. It will speed up my dev work but like you said, it isn’t a requirement. Hmmm..

 

I guess I could ask you the same question

It was the best option all in all and Lenovo took forever to get the P52 out. Have one being delivered next week though, will be interesting to compare. However, I am far from certain that the P52 will be the daily driver. But the new P1 and X1Es are interesting, direct competitors to XPS15 so Dell needs to up their game. Have not checked for actual reviews though.

 

Am I the only one btw in this thread, who has the stuck at Dell logo problem at boot and no circular loading icon appears? It is really annoying and I have no idea, whats causing it. I had the exact same issue on the first 9570 I sent back, and this second one has it too. And it is also present after a clean Windows 10 installation. I would say from a cold boot, it happens 4 out of 20 times, maybe a bit more. I then have to power it off and on again, and then it is like a dice throw if it works or not. And if Im unlucky, it happens 2-4 times in a row, and then Windows 10 boots directly into auto repair... says couldnt repair any problems, reboot, and then it maybe works again... or not. The laptop isnt frozen or anything, because the LEDs light up and shift lock press also works. It looks like a stupid Windows bug to me. Tried to disable Windows fast boot, didnt show any help for this.

 

No, had the exact same problem as reported perhaps 100 pages back. My third 9570 does not have this problem though, but a few others instead. If I recall correctly, switching back to RAID and disabling S.M.A.R.T reporting during startup led to some minor improvements, although it is hard to tell since it does not appear everytime and those machines went back anyway quite soon.

 

I had that happen to me once after I disabled a hardware feature in BIOS, it would stuck at the dell logo and go no further. After I changed back the problem disappeared.

 

Nice... I had the issue on the first 9570 with no changes to the bios and RAID with SKH SSD. I have the issue on the second one with AHCI and Samsung SSD. The only thing I changed in bios are

- enabled SMART reporting
- changed F keys to normal behavior
- disabled touch
- disabled cameras

 

New firmware out 1.4.1

Fixes & Enhancements
Fixes:
1. Fixed an issue that some systems show incorrect VGA name in BIOS setup
2. Corrected the touch screen disable behavior in BIOS setup
3. Fixed a potential system hang issue when an incorrectly formatted password is entered at the BIOS Security Manager (BIOS pre-boot password) prompt
4. Fixed a potential issue with binary image measurement using the Linux UEFI shim boot loader

Enhancements:
1. Added BIOS Password Feature: Master Password Lockout

 

Would be nice if they fixed the throttling set-point for the GPU. I get 1.0-1.2GHz in Far Cry 5 on the GPU.