9550 throttling woes making the machine almost unusable :(

Hey guys,

I'm almost at my wits end here. I'm a video editor who purchased the Dell 9550 after hearing recommendations that it was a good laptop for Premiere editors. Sadly excessive throttling is causing it to be anything but.

When I render anything out of Premiere TDP throttling kicks in after about 5 minutes and within another 5 minutes it's down to 1ghz and stays there until the encoding is stopped.

I have re-pasted and replaced my thermal pads and I have also undervolted to -1.25 for the core and cache and -0.95 for the iGPU with Throttlestop. My idle temps are about 31 degrees which seems fairly good.

So what gives? Surely 5 minutes of load isn't enough to send this machine into a complete nose dive.

I feel like I must be missing something obvious, but I can't seem to see the wood for the trees

 

Your CPU shouldn't be throttling at 82 degrees, this sounds like a power issue to me. I don't know much about that specific laptop but it sounds to me like either your VRM is overheating or your power supply is dodgy.

 

Nevermind, its a VRM issue. Take a look at this: https://www.ultrabookreview.com/14875-fix-throttling-xps-15/

Either follow that guide or try and get the laptop refunded if its not too old.

 

I'm putting a fiver on the pad replacement being an issue.

 

What kind of re***d would put mosfets and chokes right next to the CPU and GPU.And I thought Clevo is bad at designing cooling solutions, this is next level awfulness.

Yeah this is NOT a CPU problem, 82c is totally fine especially when rendering something.

 

This is an ultra thin laptop with limited cooling and limited power available from the vrm. You can do some tweaks to help but basic physics are unavoidable.

You may be hitting power limts. HWiNFO64 displays some temp/thermal/power limit throttling flags. What are they?

You are hitting all CPU cores hard as shown in ThrottleStop C0%. The vrm throttles on these at around 77*C (usually see that in 1st ambient temp sensor). Those could drive PL throttling (well thermal or PL flags are a bit vague here). Or you are hitting the roughly 45 watt limit of the CPU or the watt limit of the vrm.

Can you show all the FIVR & TPL settings on a screen shot?

One great trick to boost cooling performance is to raise the back of the computer a few centemetres; make sure you use a block of wood or rubber that is the full length of the factory black foot. Make sure you don't block the fan intake.

Also you might want to benchmark your thermals against those posted in the undervolt-repaste 9550 i7 thread. Maybe use the Prime95 & Heaven & ROG RealBench scores of others. That will give you a good idea if your repaste should be redone. Sometimes it takes a few repastes. A few repasting traps here:

1. the heatsink is sometimes warped. You can rebend it but be extremely careful as a few experienced guys broke theirs

2. What thermal pads did you replace? The VRAM pads? What did you use? If you used higher-performance pads, they are dense and if not very carefully measured will force the heatsink upwards so the heatsink makes poor contact with the chips. Fluffy low-performance pads work best for most people. [EDIT - THIS IS WHAT GONZO IS SAYING ABOVE]

3. Did you clean chips & heatsink carefully before applying thermal paste?

4. What thermal paste did you use? Is it a low-pressure paste? What application method? Did you apply the absolute min possible? Did you apply the paste, tighten the screws, remove and check quality of repaste (should be even thin film you can see good applications via web search), clean up and reapply?

Finally, there is a trick via Intel XTU to "boost" power limits I posted somewhere. That is a bit dangerous but after you exhaust all these steps, you can consider trying that out. But keep a close eye on your thermals as it could kill you laptop quickly or at least shorten its lifespan significanlty.

Study that 9550 thermals & repaste thread as everything is there.

 

That's a slightly newer version of the laptop, however I've already applied the fixes suggested for the most part. As I understand it not everyone thinks the idea of piling thermal pads up to make contact with the case is a great idea, so I've not bothered with the last suggestion on that link.

I've had the laptop for over a year now, so it's too late to send it back.

I don't think it's that. The problem was like this from purchase and has not changed after my pad replacement. The idle temps have reduced however.

The ambient readings in HWINFO64 never seem to go ridiculously high. How do I know which sensor is the VRM?

 

Virtually all 9550s had bad contact with 1 or 2 vram chips (one user had about a dozen 9550s and had the problem on every single laptop).

So you could have problems before and after the repad. Just different problems. Read my comment above

See my comment above for location of the ambient temp indicato. Your VRM is at PL limit throttling level.

 

Thanks Pressing, I'm going to try and digest your other post and will respond to that shortly.

In the mean time however there is one thing I am confused about. Why does the throttling continue to increase until the laptop is bought to it's knees, despite the temperatures all dropping considerably? TDP stays throttled until the second I remove the load. You can see from those ambient temps in my first screen shot that the temperatures are no longer high at all, but still the throttle remains.

 

Anything out of the ordinary there?

 

In the Turbo Power Limits screen, for your system I would unclick the "clamp" box for PPL.

 

Answers in bold below

 

These flags are from Intel and Dell but there is no extreme transparency and a few are clearly mislabeled. Some protection schemes will step up/step down whilst others will be on/off. Some will be driven by some combination of temps/power/time.

 

Maybe a minor issue as some heatsinks are filled with magic vapour.

https://en.wikipedia.org/wiki/Heat_pipe

That is pretty dense but maybe the 17wmk is more dense. If you made sure during fitment those were the absolute thinnist pads you could use, they did not appear to interfere with heatsink fitment on dies, your heatsink is not warped, they will not be an issue. Otherwise, it might be worth a revisit (or just put in some low performance fluffy pads).

ThrottleStop & Intel XTU do not play well together (partly because they try to write to same registers. XTU has a few major bugs also). But XTU has access to a critical third power register that ThrottleStop does not.

Search for my post when you like and and the method for getting them to work together properly (basiclly they need the same settings, they can't run together, and you need to keep XTU closed after setting up)

 

Do you think it's worth me ordering a replacement from Dell?

 

No. Also unlikely magic fluid escaped

 

Oh really? Even though it has what looks like a little kink in it?

 

A kink isn't a split.

 

Fair enough. I know they are incredibly delicate and easy to damage and I wondered whether it might be exacerbating the ambient temperature issue. I guess I'm clutching at straws here.

 

@CraftyClown Have you had this throttling issue from new or did it develop over time? If its a recent thing you should test your power supply. I had a knackered power supply on my M17X and I had terrible throttling issues with the GPUs, replacing it sorted everything.

 

Your first screenshot shows ThrottleStop PL1 lit up in red. That is your long term turbo power limit and this is usually set to the TDP of the processor which for a 6700HQ is 45 Watts. This CPU also has a TDP-down mode of 35 Watts. In your case, it looks like something on your computer has changed this power limit to about 7 Watts. ThrottleStop only has access to this power limit in one location within the CPU. Unfortunately there are other duplicate power limit registers in the CPU that can override the one that ThrottleStop has access to. This power limit can be changed on the fly. So it might start out at 45 Watts but then some sensor or a software driver decides that you are running your CPU too hard so it goes click, and changes it down to 7 Watts. This is my best guess at what is happening. Solving it is another matter.

There might be a driver on your system that is related to the Intel Dynamic Platform and Thermal Framework. This is what allows the power limit to be changed on the fly. I have not had my hands on a laptop with this issue so I am not 100% sure of what to do next. I would have a look in the Control Panel - Device Manager as well as the Task Manager for anything that looks related to this. When your laptop is idle, perhaps you can disable or uninstall this driver. I have heard that if you uninstall it that Windows will try to reinstall it. Things like that are why I like to turn Windows Updates off.

Hopefully this helps the cause and is not a dead end. At a measly 7 Watts, the rest of my hair would already be pulled out and on the ground.

 

Hey CraftyClown,
have you considered iunlock's mod for the issue of power-limit throttling?
http://forum.notebookreview.com/thr...ures-benchmarks-xps-15-9560-kaby-lake.802345/
You can try removing Intel DPTF as suggested above to check if it stops the throttling, but I wouldn't recommend this long-term because it is meant to be a thermal fuse, and the VRM area overheating appears to be real I'm afraid. It doesn't appear to be particularly easy to get heat out efficiently enough, yet possible without drilling holes in the case

 

Silly question, do you have dell power manager installed? If so, what is the fan setting?

 

Good point custom90gt. For max performance you should be running:

1. Dell Command-Power Manager 2.2.1 (Thermal Management tab -- Ultra Performance)

2. Control Panel\All Control Panel Items\Power Options (High Performance)

With the Power Options, Dell keeps changing & hiding options so you can take a look around the ThrottleStop Guide thread to figure out optimal performance.

 

It's always been an issue, although I feel as if it's gotten worse in recent months. To be honest though that could be because I've become fixated on the issue.

Thanks for the power supply suggestion. I tested with a second Dell portable supply I have and the issues are the same.

 

Thanks guys. I already have my Power options set to High Performance, however I was unfamiliar with the Del Command-Power Manager up until now. I have installed it and the default setting was optimal. I have changed it to Ultra Performance and will do another Prime 95 test. Do you think the change from optimal to Ultra performance might be a significant one?

 

Try cool setting.

Sent from my SM-G920F using Tapatalk

 

Thank you Unclewebb,

I have been starting to think that this may indeed be a Windows issue rather than a hardware issue due to the severity of the throttling and the fact it continues until the processor is at it's slowest speed despite all sensor readings declaring the temperatures are the same as if at idle. I am currently considering a full reinstall of the OS to test this theory.

I have actually taken screen grabs at significant points throughout the Prime 95 small test, from start to the processor fully throttled just 14 minutes later. Something just doesn't feel right about this.

 

Thanks Sem_. I have been considering iunlock's mod, although I still feel there is more going on that meets the eye in my particular case. In his tests his VRM temp was getting as high as 105 degrees, whereas mine might hit 80 degrees at best before quickly plummeting after heavy throttling.

I may uninstall Intel DPTF just for quick testing purposes, however I agree it shouldn't be permanently removed as it serves an important safety purpose.

Let's hope I don't end up needing to get the drill out

 

Thanks GoNz0, I may give this a try to see how it affects temps, however I'm increasingly starting to think there is something more than overheating going on here.

 

What are you testing with? My 9560 throttles 14 minutes into realbench, lol. I think it's a Dell power/temp management thing as nothing is above 75C at the time and my CPU is about 68C...

 

Yeah, something certainly seems rotten in the state of Denmark

I was running Prime95 small torture tests.

 

Humm,
I can run Prime95 without any throttling, let me make sure, I'll start another round.

 

Well 22 mins in and no throttling with Prime95. CPU temp max has been 63C and the highest ambient temp I have is 52C. I'll let it run for another hour or so and get back to you.

 

Thanks custom90gt, really appreciate you running that test, although don't worry about running it any longer as I think I may be making some progress.

For pure testing purposes I decided to fully uninstall Dell's DPTF and run Prime 95's torture tests again, if nothing more than to remove it from the equation and see if anything else was triggering the throttling.

As you can see from the screen shots below, I ran it for 15 minutes and there was no throttling whatsoever. My CPU reached a max of 70 degrees and my VRM reached 87 degrees and settled there.

I've seen much higher numbers posted for other peoples VRM sensor readings, so I'm presuming 87 degrees is ok?

The bottom line is Dell's DPTF is definitely the root of my throttling issues. I now need to find out why it is so severe.

I am going to reboot my machine now and let Windows reinstall DPTF and see if the fresh install changes anything.

 

No.

87*C is an estimated temperature near the sensor so the mosfets are running much hotter. At higher temps they get very inefficient so start to spiral out of control. Also, whilst the mosfets are likely rated significanlty higher, all your electrical components will fail at an exponentially higher rate with heat. That means early failure of your laptop.

Reinstall DPTF as it helps protect from meltdown.

ROG Real Bench is a good test for the system as it pushes the CPU and GPU.

FYI - great Miami Vice desktop graphix

 

Ok, interesting. Are you suggesting the 87 degree VRM sensor reading is infact too hot under these conditions?

I am definitely going to reinstall DPTF. I only removed it for testing purposes and to see if anything else was causing throttling.

I will give ROG a go, thanks for the suggestion.

The desktop background is actually my company logo, but yes heavily 80s inspired

 

You can run it at 87*C without DPTF. But you on average you will be reducting the lifespan of your laptop. Also what protection do have if a bunch of processes start whilst you are gaming and temps jump to 120*C?

FYI - I won't speculate what type of popular 80's merchandise your company may or may not be "distributing"!

 

What I meant was do you think 87 degrees seems like a reasonable temperature for the VRM sensor temp to have reached after about 15 minutes of a Prime 95 torture test? I have no intention of not reinstalling DPTF, I'm just trying to trouble shoot the issue.

and we're actually a video production company, so we're not distributing any 80s merchandise. I'm just a child of the 80s, so I've always had a soft spot for that kind of imagery and branding

 

Oh dear! So I allowed Windows to reinstall DPTF and then ran the Prime 95 torture test again. Worryingly things seem to have gotten considerably worse!

It took just under 60 seconds for my system to be throttled down to 800mhz. WTF is going on??

 

I ran the ROG benchmark to see how that fared and was fully throttled to 800mhz within 3 minutes

I'd love to get a few opinions here - Is this looking more like a hardware or a software issue? Sensors aren't confirming any obvious overheating issues that I can see (happy to be corrected on that though)

I feel like I have two options depending on whether it is a soft or hard issue; 1. a full reinstall of the OS if it's the former, which will be a complete pain in the arse as it is my work machine so I will need to go through all my software and back up settings etc, or 2. It's the later and I will need to call Dell out to do a service and possible motherboard replacement, which is also a nightmare as my 1 year warranty is now up, so I'm imagining it will be very expensive to have the board replaced

Feedback greatly welcomed right now

 

If it works without the driver installed then it's not a hardware issue. Which version did you just install?

 

version 8.2.10900.330 ,A03

 

Humm I have a newer version than you.
Maybe try the latest from the 9360:
https://downloads.dell.com/FOLDER04...al-Framework_MJTGY_WIN_8.3.10203.4295_A07.EXE

 

Cheers but I already tried that. Unfortunately no version past A03 is compatible with my machine and won't install. Kaby Lake only perhaps?

I am trying A02 the previous version, although designed for the earlier iteration of XPSs it might be worth seeing if it reacts differently.

 

Bummer, was hoping it would work for you. What version did you have installed before when it would throttle later on?

 

As far as I know A03 was the only version certified for the 9550. I tried A02 and although it installed ok, the results were the same and Prime95 caused full throttling down to 800mhz in just under 8 minutes.

Running out of ideas

 

And thermal management is set to ultra performance or cool in dell power manager? What about setting windows power management to high performance?

 

Thermal management is set to Ultra performance and Windows power management is set to High Performance

 

Can you access the DPTF driver in the Device Manager? What happens when you right click on it and disable it? That might keep Windows happier since you are not uninstalling it. Obviously someone had a great idea when they wrote this driver but that idea has proven to be not that great. Intel has given manufacturers the option to use a skin temperature sensor so it might be related to that. If you have access to a meat locker, it makes a great testing environment. You can also try pointing a fan at your laptop to try and keep the case cool to see if that increases the amount of time before throttling begins. If I had a spare hard drive I would definitely try installing a generic version of Windows 10. It still might go looking and install a DPTF driver. For the record, my Lenovo laptop runs just fine without needing this driver.