9550 throttling woes making the machine almost unusable :(

So I guess the next step is removing the bastard child of Satan 'DPTF' permanently from my laptop and then doing another serious stress test and seeing if it melts

 

As well install the Bloatware Free Windoze X OS from bro @Phoenix for less bloat.

You can stress test with Aida64 if you want (Less heat than P95 who is unnecessary for stress test notebooks). Two Aida setup guides. Nr.1) Is more demanding. Use what you want

1) Aida64 Stress test for more Heat.

2) Aida64 Stress test for less Heat. Should be more than enough for testing.

 

Ok, so I bit the bullet, removed DPTF and then ran the ROG realbench stress test.

I'm not going to lie, having gotten so used to being throttled if I breathed too heavily on the laptop, watching those temps soar sky high terrified me!

By 5 minutes in I had VR thermal throttling on both the CPU and GPU with the CPU package reaching an eye watering high of 98 degrees, the GPU hitting 90 degrees and the VRM sensor hitting a blistering 105 degrees.

The CPU never dropped below 2.6ghz throughout

Should I be concerned by those numbers?

 

As expected, Dell cripple the performance with software/firmware for saving a few pennies on better cooling and thermal paste!! The fans is slow to react on rapid increased load, the usual genius idea for (lower noise vs. better cooling) from Dell. Mind yooo... The Stress test is probably heavier than what yoo experience in your dayly workflow. I would re-paste with new quality thermal paste. Up to you what you want... Crippled performance ain't the solution. Same for high temp.

 

Hmmm, you'd be surprised. I work in video production so I do a lot of video rendering which is a humongous stress test.

I have just re-pasted with MX-4, so I don't think I need to do another re-paste yet.

I guess I just want to know that using my machine every day and hitting temperatures up to and above 100 degrees isn't going to damage it, now I don't have DPTF installed

 

The chips is intended for up to 100 deggees but not a good solution. And your tools (see computer) is wrong for your workflow!!

As I said... The fans is slow to react on rapid increased load, the usual genius idea for (lower noise vs. better cooling) from Dell. The average temps is the most important. Test it as it is now, in your dayly workflow. If the temp is too high. Do as suggested below...

FYI. MX-4 is Trash for Notebooks worse cooling. I would re-paste and re-pad with quality thermal paste - pads.

Edit. You could as well try with CoolerMaster U2 cooler. If you have less than 3 thumbs, maybe you could mod with better fans.

 

Out of interest how does your paste look after a test application?
Is it fully spread out and so thin you can see through it?

Repadding the VRam can cause so many issues with the CPU.

 

Pretty good I think. I always go with the less is more maxim and go for a grain of rice sized blob of paste.

When you say fully spread out, when I do a test check it doesn't always reach all four corners, but as I understand it that shouldn't be a major problem.

I greatly respect your expert opinion @GoNz0 and I would love to hear your thoughts on whether I should not be overly concerned about moving forward with temperatures in the high 90s when I do my video rendering work, or whether I'm better off accepting the throttling of DPTF if I want to continue using this machine for such tasks?

 

Several people here have had good success with the Gelid GC-Extreme on the XPS laptops. Might give you a few degrees better performance but might not. It is easy to apply and results are pretty consistent on repastes in my experience. A few degrees can make a big difference in the performance and lifespan of electronics; it also reduces the temps of the motherboard and surrounding components.

GoNz0 is really advocating people replace the 9550 VRAM pads with very fluffy low w/mk ones only when necessary (although I think all the 9550 heatsinks are defective a bad fix can be worse than the factory defect). Since you are having thermal problems after changing the VRAM pads to dense pads, I think you should buy some fluffy very w/mk thermal pads for the VRAM chips (may not be the cause but this is a popular XPS problem as the thick pads warp the heatsink and prevent proper contact causing major heat problems).

You can get the pads slightly thicker than required and they will compress easily without knocking the heatsink upwards and distorting the heatsink. They are so cheap I would buy 0,5 1,0 and 1,5 sizes for custom fitment.

 

I have tried to stay out of this as you are pushing the laptop to its limits so I would expect it to go that high, the dropping to 800mhz takes the piss a fair bit and should not happen.
It could either be a bad board (people who have had swaps have had this fixed

As this is 10 pages I don't fancy going back through as it has moved quickly in here so if I am repeating stuff you have done just ignore it.
Bad BIOS flash, always worth reflashing, trying older versions (if it works don't unfix it with BIOS updates)
A factory reset of the BIOS, use both the options, defaults and factory (can't remember exactly) and setup again, I think you can force a reload (works if you have an unbootable system via BIOS corruption) by holding the power button down for 30 seconds, the battery/charge light should start flashing when it is doing a reload and this can take up to 10 mins I think. I have had results when it has a moment and instantly resumes from sleep by holding the power button down for 5 seconds whilst it is in the diagnostics.

uninstalling DPTF is madness as it allows the laptop to run above its thermal limit, so in your line of work it will last days/weeks and go pop

I firmly believe the stock VRam pads are more than adequate as DDR4 is very cool running anyway so increasing the thermal transfer imo pumps heat from the CPU/GPU into the ram rather than it making its way down the heatpipe as it will follow the path of least resistance, this being high transfer pads, on top of this is the chance it will stop the heatsink making full contact with the CPU/GPU. I do test pastes by resting the heatsink and screwing down, running a quick benchmark (2-3 mins) to heat the paste to allow it to spread properly then removing to see how it looks. While you may think not having paste to the edge is ok that only applies to chips with a IHS attached, direct to die needs full coverage so don't worry about your non conductive paste spilling over a bit, try the X method over rice method to get it to the edge.

(I just got notification Pressing replied, to back up what he says I also use Gelid GC extreme as it works well in low-pressure applications, MX2 isn't a bad paste but needs a fair bit of weight on the heatsink to spread it out, including light rotating with the screws a couple of turns in so you can move it, maybe MX4 is the same but either way you can make it work and MX2 vs GC had next to nothing in it under load (I ignore idle temps on laptops as you can't set a fixed fan profile to benchmark between pastes))

Do your fans hit the full 5k?

Have you thought about enabling speedshift?

Have you disabled turbo?

Also, you can enable a power config setting via a registry key to limit the mhz if you want to go down that path. You should find it on the park control site.

And does the nvidia control panel use nvidia or intel GPU for the encoding program as swapping to the other may help?


If I think of anything else I will post it but work through that lot and let me know?

 

Your photos a few pages back indicate core3 and core0 temps are ~7*C different. That is somewhat chip and benchmark dependent but typically people try a repaste with 2-3*C difference among cores.

EDIT - could also indicate warped heatsink or that the hard VRAM pads are distorting the heatsink as it is a popular problem

Maybe the thermal paste will flow a bit with a few thermal cycles but I'm not too bullish. Getting that thermal issue solved might help a lot as you will dump a lot more heat out the radiator...

 

I found the 2-3 tends to be better under light benchmarking, I have an X99 chip water cooled and heavily overclocked that does 8 degrees difference under extreme benchmarks but prime95 is within 3, it is quite normal under those conditions so may relate here.

 

You have to cover the whole chips/die. If not, You will get heat problems.

What? If Dell is dependent on trash as Intel® Dynamic Platform and Thermal Framework (DPTF), so their notebooks can be used... RMA for Refund or sell it on Ebay. Also earlier and the newest Alienware models (I have one of the older ones with socket hardware) were also delivered with the same useless tech. I Uninstalled the trash and my AW17 work exactly as it did when I bought it 2013. Read also Is-it-safe-to-turn-off-DPTF

People have to STOP buying Notebooks who can't be used without need of software who Cripple the performance!! If not... This will never ever STOP!! Vote with your wallets. Maybe you will see a change.

 

The trouble is some people expect an ultrabook with a dedicated GPU and a quite high wattage GPU to run like a PC, at least an Alienware has decent cooling so you can get away with removing the limits.
Yes it has problems with overheating the motherboard but I knew what I was getting into, I may do some light encoding with handbrake and it runs like a champ with a repaste, I mainly got this to take on holiday for some light singleplayer gaming when the kids are asleep.

FYI I did vote with my wallet at the time as I couldn't find anything remotely close back then. If Toshiba had a solution close to this back when I got my 9530 I would have gone Toshiba as I worked for them at the time. Due to all the problems I had with it I got a full refund 35 months into my 3 year warranty (having already been upgraded to a 9550) and for the price went with a 9560.

I probably wouldn't if I had £1800 in my hand today!

It shouldn't drop to 800mhz though, something is wrong for it to stick at that.

 

People should expect every single sold Notebooks can be used fully within Intel and Nvidia's own spec sheets.
I can't understand that people havent read or have the knowledge what Intel ® Dynamic Platform and Thermal Framework (Intel® DPTF) really is. The Warning bells should have begun calling/ring for long time ago. This is not a new tech!!

 

After a good paste job, CPU and GPU should be in the eighties or lower in stress tests. CPU near 100 is not normal. Only PL throttling after a few minutes is unfortunately the norm with the 9550. Kryonaut and Gelid EC are most recommended, but MX-4 is probably okay. But more likely you've got an alignment issue.

Removing DPTF may be acceptable when you're sure your laptop is running without issues and you've done at least something extra about the excess VRM heat (undervolted, padded the mosfets...). I wouldn't do it in your place (yet) because you've got issues.

 

I am considering ordering some Thermal Grizzly Kryonaut paste as it too seems to be very highly regarded. Regarding thermal pads are there any specific ones you would recommend? I've seen a few people mentioning to go for fluffy low w/mk ones but I'm not really sure what I'm looking for.

 

Thanks for the detailed response GoNz0, very much appreciated.

So to answer your questions;

Bad BIOS flash, always worth reflashing, trying older versions

I've tried two Bios versions as well as full resets of bios settings, so fairly confident the answer doesn't lie there.

uninstalling DPTF is madness as it allows the laptop to run above its thermal limit, so in your line of work it will last days/weeks and go pop

Yes, sadly I agree with this. DPTF is hugely annoying and massive overkill, but it doesn't change the fact there is an issue with my device and despite being an irritation it may currently be the only thing standing between my laptop and a complete meltdown... Literally!

I firmly believe the stock VRam pads are more than adequate

I swapped the original Dell ones out a while back, but I am strongly considering finding some more spongy/pliable ones to use in place of the Fujipoly ones I am currently using, however I am struggling to find any specific pads that people recommend.

die needs full coverage so don't worry about your non conductive paste spilling over a bit, try the X method over rice method to get it to the edge.

I am going to try the X method the second the new Kryonaut paste I have ordered arrives.

Do your fans hit the full 5k?

Yes the do. I have them set to Ultra performance.

Have you thought about enabling speedshift?

I am using Speedshift and it turns out that having my EPP set to either 50% or 65% is the reason that my throttling goes immediately to 800mhz rather than gradually down in stages. If I set EPP to 0 I still end up at 800mhz after a few minutes, just not quite so quickly

Have you disabled turbo?

Yes I did also test this straight away, however I still throttled quickly

Also, you can enable a power config setting via a registry key to limit the mhz if you want to go down that path. You should find it on the park control site.

I will bear that in mind if I can't get this fixed any other way

And does the nvidia control panel use nvidia or intel GPU for the encoding program as swapping to the other may help?

The software I use is optimised for Nvidia GPUs, so using the iGPU would defeat the point as it would be as if the machine was throttled, besides it's a moot point as both options still result in quick throttling.

If I think of anything else I will post it but work through that lot and let me know?

Thanks once again for your help so far

 

The Kryonaut is quite popular for XPS users.

The original Dell pads are fine from both thermal and fluffiness specs (as long as they make good complete contact between with both heatsink and the VRAM chips).

I have some non-branded 4 wmk pads and the Fujipoly 17 wmk pads but that does not help you much. Do a search here as there are a few recommendations in the XPS threads. One guy used the GELid 7wmk pads if I remember correctly. Low rated pads tend to be fluffy and probably are sourced from just a few of the same suppliers

If you like Fujipoly, see p.15 for a graph comparing compression vs thermal conductivity. The very high performance pads tend to be very dense as you see but say 4-5 wmk pads are less dense and specific models you can choose.

https://www.fujipoly.com/usa/assets/files/Fujipoly 2016 Sarcon Catalog for web.pdf.pdf#page=15

 

I quite agree. Once I know the machine is not heating like a furnace under load then I will more seriously consider retiring DPTF

 

So it appears the most compressible Fujipoly pads are the XR-Ms but they are 17w/mk. Am I looking for something compressible but ideally lower in w/mk?

 

Have already posted solutions for thermal pads for you in post #106. Arctic Cooling thermal pads is soft, cheap and well suited. Don't go for the stiffer and more expencive 17w/mk pads.

 

Use the pads Papusan recommends.

There are no reported problems with pads in the 4wmk range for the VRAM chips as long as they are making full contact to both the VRAM chips & the heatsink

 

Ah yes, my mistake. I went to the thread and took recommendations on Thermal paste, but didn't spot the section on pads. Thanks again

 

Are these the ones we are talking about here?

https://www.amazon.co.uk/gp/product/B00UYTTXSM/ref=s9_dcacsd_dcoop_bw_c_x_1_w?th=1
https://www.amazon.co.uk/gp/product/B00UYTTLI4/ref=s9_dcacsd_dcoop_bw_c_x_1_w?th=1

 

no need to spend that much, any old cheap pad will do.

Sent from my SM-G920F using Tapatalk

 

Alienware's too have issues with overheating.

For thermal pads choose arctic because they are soft.
Check this link out, do check how improper paste application looks (First 3 images and fourth is the perfect one and has great temps)
*OFFICIAL* Alienware 15 R1/R2 Owner's Lounge

 

With an additional £300, you now have £1700 sunk into this, correct?

I would not be doing the 'tests' you indicate in post #103. Instead, I would just be using the notebook as you need it: i.e. in your actual workflows. What are the temps then? If they're as ugly as that previous post indicated; I'd be selling the notebook, complete with the new 'warranty' that Dell happily sold you (I don't think they did you a favor offering that...).

There is some great advice from many here to help you try and fix the problem. I'm in the camp that if it doesn't work out of the box as intended/expected; it is returned/sold/gifted asap.

Even if you get an additional 2 years out of this system - which will seem like torture to me with the system as described - for your intended workflows, it seems like it will be pushing it to achieve even 70% of what is should be capable of, spec-wise - now is the time for a new system (when this one is worth something...) - not after you've spent more time and $$$ on achieving (much) less than what a new platform will give you in spades.

I understand there are other aspects to consider from your personal perspective too - but this post highlights my thought process and my black and white solution.

On the other hand; with DPTF removed - using the notebook as intended (i.e. your workflows) - and with temps in the somewhat reasonable range - I would use the notebook till it died - you do have 2 years warranty on it after all.

Hope to see this story turn positive for you soon!

Take care.

 

Well the way I see it is I already had the £1400 invested into a machine that had great write ups from a number of video editors, so it seems that when working correctly it is capable of doing the job I require of it.

Now I have the warranty renewed I will have another crack at repasting and repadding, but if that doesn't help then I will get Dell to change out any necessary components (I'm guessing motherboard) until it does

I guess what I'm saying is that rather than wasting money, I'm trying to ensure I get my money's worth.

Oh and I definitely will work it to death now I have that extra two years.

 

To clarify; there is no 'investment' part of owning a notebook. It is a sunk cost. You gave the $$$$ or otherwise paid for it a long time ago. What you do today going forward should have nothing to do with what you've already done w/regards to this expense.

See:
https://youarenotsosmart.com/2011/03/25/the-sunk-cost-fallacy/

Regardless of how many others praised this system; yours is proof that your sample is anything but capable for your workflows (as-is, but even 'fixed' I wouldn't trust it for the long term - the design is faulty; replacing parts indefinitely is not an option (for me)).

The short of it is that "you've already wasted the money". Don't continue doing so in an emotional fashion to keep the same substandard (for your workflows) system/platform.

The new options you have today are far better than when you bought that platform almost 16 months ago. Assuming the budget was there; I would be jumping at the chance to upgrade from the system you presently describe.

At the most; I would replace the M/B if needed under warranty and then immediately get rid of the system.

(You have an almost identical experience of a client of mine from a decade ago... I gave the same advice then too - he had the system 3 years and got less than 18 months worth of use out of it - the rest of the time it was having one issue or another and he would be waiting for parts and the Dell tech to arrive and take his system apart once again - in the end he sold it for parts... if I remember correctly? He told me he made $125 on those parts - on a $2K machine).


With the proceeds put towards a new platform, I am certain that (from many past experiences...) that will be the best use of funds in the long term.

Good luck.

 

I would hope either they have no mainboards or can't resolve it as that will result in you getting a 9560!

Sent from my SM-G920F using Tapatalk

 

Oooh. I hadn't even considered that possibility!

 

Yeah having a recurring problem a few months before the warranty is up has its perks. Once the outlet no longer stocks your laptop you can be sure parts are also load stock, they have to upgrade you then.

Sent from my SM-G920F using Tapatalk

 

Good afternoon all,

I thought I should update this thread with my progress as it's been a couple of weeks.

So as a brief reminder I have the 9550 6700HQ with 4k screen, 16 gig of ram and the 1TB NVMe PM951. I occasionally use this machine for my work which puts the machine under quite a bit of load as my business is video production meaning both the CPU and GPU can be heavily taxed for long sessions.

I started this thread as I had been experiencing severe power limit throttling within minutes, under what I would consider medium to heavy loads. Upon testing I discovered I couldn't even get through a Prime95 test without being throttled down to 800mhz in less time than it would take to make a cup of tea.

Fast forward a couple of weeks and I have got this puppy purring like a kitten now. Not only can it cruise through a Prime95 test, but I can also finish a 15 minute Realbench stress test without the throttle kicking in. (Disclaimer; This was with the device on a basic cooling pad)

It's taken a fair amount of blood, sweat and tears to get here and a great deal of what I tried made no difference at all, but looking at the performance improvements I'm getting I'm glad I made the effort

So my current setup is as follows;

1. Targus AWE55EU cooling pad. It's not particularly powerful, however it does reduce temperatures across the board by 1 or 2 degrees whilst also allowing me to work more comfortably.

2. I have added thermal pads to the VRM chokes. I used 2 x 1mm thick 6 W/mk pads to ensure they make contact with the case back. This has reduced the VRM ambient temps considerably and I rarely now see them rise above 80 degrees, even under sustained heavy loads. Despite the pads making contact with the case, their low conductivity is preventing the back side of the device from becoming noticeably hot, even without the use of the cooling pad.

3. Using Throttlestop I have undervolted my CPU Core and Cache to 150.4 mV and the Intel GPU to 99.6. My system appears to be fully stable with these numbers.

4. Once again using Throttlestop I have set Speed Shift EPP to 72. After much testing this is appears to be a sweet spot where I can maintain maximum speeds during the Realbench stress test.

5. I have re-pasted the CPU and dGPU with Thermal Grizzly Kryonaut paste using the rice sized application method.

6. I have re-padded the VRAM with 0.5mm 6 W/mk pads, except for the chip below the heatsink pipe where I have used a 1mm pad to ensure contact.

7. I have updated DPTF to version A08

8. My bios is 1.14

9. The Intel GPU is set to Max battery

10. The Nvidia GPU is set to High performance

11. In Dell Command power management I have thermal management set to Ultra performance

12. Under power options I have high performance selected


Some of these tweaks and mods have probably made a lot more of a difference than others, however the bottom line is that I am now able to use the device as I had originally intended, without fear of a complete meltdown.

Many thanks to all those who threw their advice into the pot and I sincerely hope this thread proves helpful to any others who may have similar issues.

 

Thanks for the update @CraftyClown

You are one of the few that have balanced the case bottom cooling challenge. It is easier with the 9550 as there is less overall heat than with the 9560. Plus you are using an external fan & weak thermal pads on the chokes only.

You might be able to tweak a bit more performance in ThrottleStop (See @Vasudev comments in Throttlstop Guide thread):

- Look in TPL window (reallocate Intel Power balance between Intel CPU & Intel GPU). My 6300HQ is at 31 and 13, respectively. I notice the bump in speed but additional lagginess of the Intel GPU during mundane tasks.

- You might not want to underclock the Intel GPU as it seems to cause some studdering.

- See @unclewebb 's comments relating to getting your C-states in C7+ say 98%+ of the time during idle to minimize thermals. That requires a bit of tweaking but the additional thermal headroom helps

Also, you might consider bumping up the package power limits and times a bit (you need to do this both in Intel XTU and ThrottleStop- search my threads for details and warnings).

 

Thanks

Regarding your additional suggested tweaks;

1. How is the Intel power balance worked out? By default the 6700HQ appears to have it distributed as 9 to the CPU and 13 to the GPU. Is there any other info or documentation on this?

2. I've not heard anyone else mention to not underclock the iGPU and I haven't seen any stuttering. Is this a known issue?

3. Cool. I'll take a look at Unclewebb's comments on C7 states.

4. Would increasing package power limits not push me back in the other direction towards PL throttling?

 

Q1:No documentation just experience. You can tweak it as you like. Under CPU heavy tasks setting more power to CPU always help because preference is given to CPU instead of iGPU.
Q2: Only handful of people notice this on Windows UI stutter when iGPU is undervolted. Even web browsers lag sometimes because of undervolting iGPU. Like I said, only small group of people noticed this(myself included).

 

This helped me reduce latency performance on my computer for huge live music software. #1 & #2 provided a big improvement on my system, which is already optimized for ultra-low latency.

Search for my postings if you are bumping into what are flagged as power limit issues (PL). It might give you a bit more headroom at the margin. But if you are pushing hard, there is only so much power the VRM will spit out and it will eventually run out of steam.

 

So if I understand correctly you're saying removing the undervolting on the iGPU prevents stuttering and latency issues?

 

I did #1 and #2 and low latency performance improved for my Virtual Instruments. I did not test those independently. YMMV but give them a try. . .

 

Hey CraftyClown, and everyone in this discussion!

Thanks for posting about this issue and congrats on getting your system working!

I've just received the 9550 refurbished from Dell outlet - and am in a similar situation, though it seems even worse/quicker than yours, the throttling kicks in immediately. Doesn't go over 7 watts...

I am curious as to your stated:

DPTF Version A08 (where did you get this from?)

and

BIOS 1.14

I have just installed what seem to be the latest for each, being the below:

Dell XPS 15 9550 1.5.1 System BIOS
Intel Dynamic Platform and Thermal Framework / Version 8.2.10900.330, A03
Release date : 06 Jul 2016 / Last Updated : 08 Feb 2017

But the felt no gains.

If you could take a minute to show me how to get the versions you used I'd appreciate it.

I also wish to edit video, currently this is slower than my XPS 13, I get 800Mhz instantly and can export 4k at 1 fps.. and the temperature is 40-50 degrees. Both CORE and GPU PL1s light up red straight away.

Thanks in advance!

 

use a stand and see if it helps in reducing temps.

 

Send it back, probably why it got returned the 1st time.

Sent from my SM-G920F using Tapatalk

 

Thanks guys.
I don't think it's cooling as it doesn't go over 45 degrees.

I disabled the dptf stuff and got to see that the cpu does go over 30% utilisation and 0.8ghz. But the temp went to 97 so I stopped the render.

Going to send it back. Shame.

Thanks again.

 

@unclewebb I've been reading all the posts about using Throttlestop for the i7-6700HQ and so far I've been getting much better temps when doing isolated CPU stress tests with a 150mV undervolt. However, whenever I run a simulataneous GPU and CPU stress (to simulate gaming) or play a game for an extended period of time, I get TDP throttling that kills my performance. I haven't been able to find a good explanation about what TDP throttling actually means and how to fix it. Can you help?

P.S. I haven't opened the case and repasted or added thermal pads yet (all the supplies are in the mail).

 

http://forum.notebookreview.com/threads/tdp-and-power-limiting-haswell.766743/

 

So I read through that post and I don't quite understand what the solution to my problem is. When playing games I get PL1 throttling in both my "core" and "gpu" as well as EDP OTHER throttling under "ring" in throttlestop. Would adjusting the package power limits or primary plan power limits help? If so, by how much and is there risk involved?

 

Risk is involved as you are running more power through system than expected so can shorten lifespan and/or kill system. Or brick it with RW-Everything. Etc...

I run package power limits at 65 watts with no issues. PP0 current limit of 90. Sometimes need to reset.

If you increase certain factors too much, system will ignore your requests.

In ThrottleStop (turbo power limits) you can set package power limits and primary plane power limits/time. I think ThrottleStop does not have access to some registers including a MSR and maybe more. Give it a try.

Alternatively you can try Intel XTU, which is very buggy but seems to access all the registers; you need to check Intel XTU constantly as it has a tendency to randomly reset or change your settings. Note you can NOT run ThrottleStop and Intel XTU at the same time. You should have identical settings in both programs so if you swap programs things don't get confused.

You could also run ThrottleStop & run RW-Everything (everytime you start to access one additional register which seems to work for my system).

I have some posts you can search for my 9550 6300HQ. Also, during the past few months there are some details on the dedicated ThrottleStop thread.

Also people are using MSI Afterburner to adjust GPU on 9560; look at that thread as it might work on 9550 also.

 

6700HQ have locked down PL1 power limit (45W). You can't increase that Power limit for your chips.

 

Correct - I should clarify. The 9550 vrm has zero headroom to boost current in any case.

But....my trick does something to reduce the notorious PL throttling in the 9550. It is clearly fooling something in the system allowing benchmarks and games without PL throttling. This was more "luck" from tinkering than any exciting "discovery".

The most obvious improvement was stable CPU speeds using MSI RealBench (which kills this laptop as it tests both CPU and GPU).

That is with a 6300HQ; your 6700HQ runs much hotter so you might not have as much "luck".

The other "trick" with the 9550 and 9560 is that Dell purposely throttles when the vrm gets above about 76*C (PL throttle interestingly). I think Dell did that because the vrm is so undersized. Regardless, if you can keep temps there lower than about 76*C, that removes another route to PL throttling. (labeled as one of the ambient sensors in HWiNFO64).