XPS 15 9570 Owners Thread

@maffle I don't have major power drops with modern standby. Everything is green for me HW100/SW100. Still a rate of 650mw.

Have you guys solved the Intel Audio service problem and the Abnormal shutdown notification in the logs?

 

Apple seems to have misconfigured power limits in the new i9 MBP:
https://www.notebookcheck.net/Throt...improperly-set-VRM-power-limits.318176.0.html

Has anybody checked if this is done okay in the i9 9570?

 

Related:

https://www.notebookcheck.net/Apple...l-throttling-in-new-MacBook-Pro.318232.0.html

 

Is it just me or have the quality of Dells software improved in the last year? Used to be that programs like Dell Updater seemed like something an intern throw togheter in VB6 10 years ago but now the seem to at least not s-u-c-k as much as before. The new Support Assist software, when it works at least, isn't to bad either. And as frequent visitors to these boards might remember, I have not always been to happy about the Dells software

Download page is still s-h-i-t though...

 

All you can do is undervolt. It's Dell you know...

Dell XPS 15 9570: 15 % more performance by undervolting

 

The VRM modules have been changed/upgraded since the last year model.
I am not an expert, but I never run into the problem of having them not provide enough juice to CPU/GPU and cause throttling.
They do get hot under stress test though.

 

Thankfully with Dell's decision to limit package power the VRMs don't have to work as hard, lol.

 

Whats going on with the fan curves using the latest BIOS? I'm running Prime95, and the fact that there is throttling does not surprise, but rather the fact that the fans go up and down despite the CPU load being 100% constantly. Could it be the Power Limit throttling causing this? No thermal throttling reported by HWInfo (probably because Power Limit throttling occurs before).

Speaking of HWInfo, I can't seem to find any info regarding fan speeds anymore in HWInfo, has Dell changes some chips or something so HWInfo cant read it? Quite sure I have seen this on the 9560 in HWInfo...

 

HWinfo shows the fan speeds, either you managed to turn the sensor off or you're just missing it.

 

@custom90gt DIMM sensor threshold aside.
Did you know that the razor blade 15 is hard configured to 35W always?

@improwise look better.

 

Had to reinstall HWInfo, for some reason it just was not working any more. Or rather, no data was displayed at all in the section where it was supposed to be...

 

I had heard that, but I also heard there was a fix to make it run at 45w.

 

Well they still don't have active cooling while they do get jolly hot, 125 degC at the first Ambient sensor has been reported recently.

Though, the specific question here is if the power limit set by Dell matches what the VRM can feed. I understand the Apple problem was that they did not reduce the limit to the actual VRM capability.

> "... Doing the math, the OP says that Apple has shipped the Core i9 MacBook Pro with CPU regulation power (Power_Limit_2) set at 125W for 28 seconds and Steady State limit (Power_Limit_1) set at 100W. The problem here is that these are Intel recommended settings for this processor, which assumes that you are able to supply 100W for the steady state. Testing by the OP has revealed that the VRM can supply up to 49W before maxing out (not counting the GPU). Since the VRM is not able to supply the required steady power limit due to overheating, the motherboard signals the CPU to slow down to 800 MHz to allow the VRMs to recuperate. ..."

 

And the answer to that question is yes. For the XPS15 i7, the power limits are: short-term 75W, long-term 56W. The VRMs can do that. I don't get throttling because of inadequate VRM modules.

Under stress test (both cpu and gpu fired), I get first ambient up to 105c which is hot but well within spec (I checked the manufacturer page).
Would I prefer cooler mosfets? Yes, of course.

 

Damn... I just get SW100 / 25-50 HW, but not 100 100. Can you post a sleepstudy report sceenshot, please? Mine always show some "c7 stuck" that is what I interpret of it. How much total battery percentage drain is 650mw for you per hour?

I could actually "solve" the Intel Audio Service crash and the abnormal "Windows didnt properly shut down last time" log entries, by deactivating the Intel Audio Service service under Windows services. This also removes the "Windows didnt properly shut down last time" issue, so they seem to be connected. I have no idea what the Intel Audio Service actually does or what it is good for. Audio works fine so far without it running. No idea, why some people dont get it. I got it on two 9570 which came fresh from Dell.

I also read your DIMM throttle fix post, by "faking" its temps of shielding it from the heat pipe. Is there maybe a better how to for this somewhere with pictures?

Also... could it maybe be done, that you put thermal pads on the VRM area so it connects with the aluminum back chassis? The bad thing is the chassis has this stupid "thermal forceshield" attachted ot it at this position, you would have to remove that first I guess.

As I see it, wouldnt it also work to connect the heat pipe area with the chassis? this could dramatically drop temps. The back chassis may become a bit warmer... but.

 

Are your power limits accurate?
According to my logs it's 56W for around 2.5minutes and then 45w long term.

And I don't think you can say that it's not throttling because of VRM when you haven't looked at any logs from 120+C VRM systems. Mine didn't run anywhere near that, but if it did I would be curious what the logs looked like. I'm sure they have VRM throttling at some specified temp, but because mine didn't get that hot I never ran into it. I don't think you can speak for the others unless you have logs where your VRM temps are hitting 120C...

 

I have a 9570 on the way and want to prepare for getting the heat under control. My plan is this:

Undervolt CPU and GPU
repaste with Thermal Grizzly Kryonaut (I've always used Arctic Silver 5 but this stuff seems to be the best currently)
Install 6 copper heatsinks w/ adhesive thermal pads on top of FETs as shown in this image: http://i.imgur.com/sjHJNSR.jpg

I found that pic in a thread about the 9560 and it was combined with a method to channel air flow, but that's really more drastic of a project than I'm willing to take on. What do you guys think about this plan? After driver/bios updates, I plan on benchmarking/reading temps completely stock, and then doing the same with the mods.

 

Give it a try. Custom90GT has a thermals thread so that will give you some answers before you start. . .

 

This is for a 2h20m sleep. I have nevertheless contacted Alan about the S3 bug and he told that they have escalated the problem. We'll see.

I did some digging and discovered that this Intel Audio Service is not related to playing anything. It's not the Realtek Audio driver and It's not the Intel's HDMI output.
This is a server/driver associated with an integrated DSP in new Intel processors to speed up voice recognition (Cortana, etc): intel smart sound technology. For whatever reason, the driver for this service is bundled with the Realtek's drivers on Dell's website. I tried to upgrade this driver both with dell's provided package and using device manager "updated driver" feature. We'll see how it goes. If it doesn't work, I'll notify dell support.

You can either: 1) pad the (top) DIMM modules with 1.5mm pad, 2) pad the sensor with a stack of 3x1.5mm pads, or 3) do both.
I did both but my personal option is that 2) is just fine in this laptop.

So you can do that. I was actually planning on doing that and I did some tests. Pad every VRM module to the backplate, you need 3x1.5mm pads (or equivalent).
The problem is the back chassis gets a LOT warmer, toasty! Of course, we are talking of heavy stress tests here.
On the positive side, it means that padding works for transferring heat, and my ambient1 was hovering in the mid-90s.
On the negative side, the back plate becoming so hot makes me believe that it is not a good "heatsink". It doesn't have fins nor fresh air blows on its surface (unless you use a cooling laptop pad). @pressing was speculating that it might get so hot that it ends up heating the intake air doing more harm than good (this is a very sensible observation!)

You can use tiny heatsinks instead of thermal pads, but that should be more or less the same.
You can use tiny heatsinks AND try to divert some air from the fans to the VRM area, like @custom90gt. This may be the best solution (and most laborious one).
However, in his test, he got better results by modding only the MOSFETs in proximity of the sensor. That doesn't mean that the other MOSFETs don't get hot.
These things are very hard to assess without hours of A/B testing and extra equipment.

There are other layers in this discussion. The price, the TIME, and the skills involved in implementing and testing these mods.
People say they are ok modding but after you spend 3 days trying things and running benchmarks, you start asking yourself why the hell should I do this on a new laptop (ask @Woodking).
Another layer is the RISK associated with each mod.
What if you strip the battery connector during the tens of mounting and unmounting?
What if you use the tiny heatsinks and one day you shake the laptop a bit harder in the backpack and one of the heatsinks gets dislodged and creates a shortcut and you fry your motherboard? (And yes, you can use heatsink+pad, but a risk is still there)

At the end of the day, it's your call. Based on what I read and my first-hand experience, my call was to not mod the VRMs.
I don't think it's *needed* to obtain better performance (as opposed to 9560) and the risk/benefits aren't worth it (to me).
I might change my mind in new results come out.

PL2 is 75W for about 30 secs. PL1 (which should be the long-term boost power limit) has a default value of 56. Digging in my logs I have seen it go down to 45W the "TDP" of the 8750H due to a variety of factors that I don't fully understand. I don't think it's time or it's just about time.
This PL1 value is clearly operated by Dell's power management and we should reverse engineer what all the factors are (DIMM sensor 62c threshold being one of them).

Yours is a fair argument. All I can state is that I haven't seen throttling at the temperatures I reach for ambient1, i.e. 105c under prime+heaven. And I think my temperature didn't go higher because I undervolted CPU and GPU. Can there be a threshold at 120c that I didn't see? Sure. I recall one user here in the forum lamenting 120c degrees under stress, but not crashes. It would be nice if people who have these issues posted here along with some temp measurements so that we can figure out what is going on.

 

You're right, it's not just about time. Looking at 5 random logs here is the timing of my 56w to 45w PL1 throttling:
01:59.9
01:58.3
01:51.6
02:11.0
02:17.1

The last log was my prime95 run with all of the mods in place.

Maybe when Dell releases the 9570 on the outlet for a decent price I'll pick one up again and look at it more. Although I do have 94 logs that I could sift through, ehh what's the fun in that, lol.

 

I got a 9570 unit with a fan that sounds like a moped so that's nice

 

@abujafar

I spoke about the Intel Audio Service crash too with the Alan guy from Dell So you can tell him actually too you have the issue. He said he can't reproduce it (of course... typical). I am not sure if he escalated that ticket too or not.

Can you put the laptop sleep longer than 2h? Maybe 12h overnight in modern standby? And see if you see have 100 100?

Do you have a picture maybe of before/after with the DIMM sensor fix?

I see... I was thinking the same if it maybe would more harm than good. I won't bother having the chassis back be warm actually, I never have it on my lap. I also don't really stress the laptop 100% lots of times. It may actually do lots of wonders for idle temps and mild gaming with caped 60fps and low settings, so the fans won't kick in anymore for browsing and video watching.

Do you maybe know, which components in the middle top voltage area are the coils? Is there a picture where they are marked? I read somewhere before, that you can actually use glue (the right one of course) you put around/on the coils and enclosure them by this like amber, to deactivate coil whine. Would that maybe work?

 

I'm gonna give it another try. Dell/UPS informed me new 4k 9570 will arrive today.
I have no questions about the mod/undervolt/repast stuff ( i did it already many times before, 9570 and other machines) but i think i will do a fresh install on AHCI and preconfigured "old" S3 mode. Any advices?

 

I don't think there is a limit - I guess it would have been much lower if it were set (in was in the seventies on the previous models), and then the laptop would have throttled more of course. But you can look up other folks' report at Dell reddit and their community support forum - up to 125 reported, and others don't measure temperatures but report freezing/chrashes under load.

 

The problem is that there is zero airflow inthere, so the heatsinks alone don't help much. It should be better if you funnel a bit of the cooling air over there with thermal pads, and it isn't such a big deal. Obviously iunlock was trying many simpler mod versions and he wasn't happy with what he got with less involved versions - he was after unthrottled performace. You may or may not be satisfied with the outcome - you can still try harder if you are not. Considering the difference in the 9570 throttling scheme, consider padding the RAM (incl its sensor) to the backplate. You must do something about VRM heat, but don't pad VRMs to the backplate.

 

My hinge makes a popping sound like it's loose but the lid is really stiff and not easy to open, I've gone in and tighted all the screws on the hinge and still does this. Any ideas? Or should I return the unit?

 

Left in sleep mode overnight. Still 100%/100%. Perhaps you have some process that wakes the system up?

There are some pictures of the laptop in @custom90gt post by him and others.
If I open the laptop again, I'll take a picture.

Do your experiments and report back.

I don't know to fix coil whine. I was under the impression is that is something that can "randomly" happen for a variety of factors and manufacturers haven't figured out a systematic solution.

 

You know, at a certain point I had a theory for explaining this behavior. I noticed that in LONG TERM CPU stress tests the CPU tends to sit in the low 80s. Perhaps, Dell's power management adjusts the PL1 to achieve this target.

 

I have SW 100% and HW 5% on connected standby, and it's because of some NoHwDrips.C7 on SoC which was active 95% of time. 4% battery drain for 1:46 hours. This was with wifi and BT on, and with Chrome, email and few Office apps opened.
I'll do a 10 hour test now after this, but I'll put the laptop to airplane mode before I close the lid.

Since your laptop manages to sleep nicely, can you please give us some details, like if your wifi and BT were on during the standby, and if Chrome was open?
Have you uninstalled any Dell apps that come installed with the laptop?
Any registry or Windows settings changed?

I think it would be REALLY helpful to get to the bottom of this sleep thing - being able to put the laptop to sleep without worries instead of using hibernation every day is actually a huge deal for daily practical use.

 

Wifi ON, Bluetooth off.
However, powercfg /a says that only "Modern Standby Network Disconnect" is supported on my device.
In this case, the wifi status *should* not matter.

I had almost all the applications closed. I did not remove any DELL app.

Regardless, I do hope that Dell fixes the old standby (S3).

 

Had the same popping sound issue with the FHD unit I returned.

Additionally, did anyone else have hissing through headphones.

 

That is totally the SAME on mine, what I had reported with the "NoHwDrips.C7" drain. Not sure what it actually means. Does it mean, that it never went past C7, so it never could enter C8-C10, or does it mean something else?

I had closed all apps of course for the standby test, no Chrome, no systray apps, nothing, literally nothing running. I have also nothing really installed, no services or whatever. Had double checked via Task Manager, 0.1% idle. I had uninstalled all Dell apps but the Power one of course too long ago. (That bad Dell System Analysis app, based on PC Doctor, had app crashes too all the time in logs). The only thing running is actually ThrottleStop. I just used it for undervolting, nothing else.

There is actually no option anymore in Windows 10, at least on the XPS, to turn off WiFi during standby, can you activate it somehow via powercfg? I totally dont need and want to have WiFi and BT running during modern standby. Actually modern standby just does not make any sense on a laptop. I told that the Alan Dell guy too, because he asked me, why I want to use S3... and he didnt understand why I dont like modern standby...

 

How can Dell make any money on these machines with all the returns? Just requested a second return now, both times due to stuck pixels in the center part of the screen (or not dead but stuck). Unbelievably...

Considering the ZBook G5 studio now, anyone else looked in on that machine? Also rumours that Lenovo will finally release a direct XPS competitor...

Throttling I can live with (or rather knew of before purchasing) but these constant QC prolems drive me crazy

Sent from my SM-N950F using Tapatalk

 

Yup it's crazy, seems to be common sense by just putting a little money into quality control and it would save them all the hastle and dollars of returns. I've seen this with multiple companies so we can't blame it on Dell. I've seen it on Asus, msi, Razer and hp. Now Dell so yeah it's all of them lol can't say for Lenovo though

 

Returned it and got a new one without this hinge issue, happy with the new device so far!

 

EDITED
Some infos about the Lenovo X1 Extreme (in German).

http://winfuture.de/news,104274.html

 

That entire video only showed the X1C6. Even Lenovo isn't that reckless...

right?

fellows?

 

@jeremyshaw I think you are right! I really don't speak german...
I edited my original post.

 

removed

 

bought this ram and it wont boot even though i checked the crucial website to confirm its compatible, WTF. TIME TO RETURN A HOLES

https://www.amazon.com/Ballistix-16...i=B06XRBS4Y5&refRID=HQHTYBA97PCV0N3GG350&th=1

 

I think other users have already reported that that RAM is not compatible in this thread or elsewhere.

Check these links:
https://www.reddit.com/r/Dell/comments/8r55ys/9570_working_ram_kits/
https://twitter.com/AzorFrank/status/1009455354076332033?s=20

 

yeah ive found that out now, reason im returning. crucial states on their website they do support it so i took their word for it before clicking buy. i should of known new devices encounter these issues. no time to wait. crucial shoudnt have that up on their site yet. lesson learned time to return

 

The amount of times I've told Crucial this yet they still say it is compatible, but they admitted to me on multiple calls that they hadn't actually tried it in a physical machine - so how can they say it works and is compatible then? The website should be updated if its proven NOT to work.

Crucial CT16G4SFD8266 16 GB (DDR4, 2666 MT/s, PC4-21300, DR x8, SODIMM, 260-Pin) Memory is compatible, in either slot but its not in the Ballistix range and is CL19.

 

How long did it take for you to get a refund, I've been chasing them after mine was collected for almost 2 weeks now and keep getting passed from person to person as they all seem to be going on holiday or off sick?

 

They will return just after your refund window closes. . .

 

Have not askes for one yet, but usually 2 weeks I think

 

His machine is already collected.

 

Hmm, likely I guess. Good job I documented everything and have CCTV footage of the laptop being collected. I think they try to hold on to the refund as long as they can, and if its much longer then in that case they'll lose another loyal customer for over 20 years.

 

... I just love it, I just LOVE it. This is the reason why I hate connected/modern standby. IT NEVER WORKED on ANY device I had in the past. There must be a way to find out, whats causing it. I even deactivated every "fishy" service with msconfig... no help. No program is running.

 

This one looks nice.