The ThrottleStop Guide

Last question. Is DTS is just the alarm? or will it shut down the pc etc..?

 

@unclewebb you should see this:
PowerCut disabled and the CPU is reporting what seems like the REAL power consumption. I am not sure this is due to the microcode update, or something else caused it. As you may remember, before I wasn't able to go beyond TDP no matter what.
Mysterious are the ways of the CPU...



Even though lately I didn't get any BSOD, when my laptop recovers to sleep, PowerCut is disabled even if VCCIN was not set to default.
As I recall, the instructions for disabling PowerCut were:
1. reset VCCIN to default
2. put laptop to sleep
3. recover from sleep - and PowerCut will be disabled.
In my case, it seems PowerCut disables every time when recovering from sleep.

I also noticed higher temperature. Yesterday I ran the CPU on full for approx 5 hours and the temperature was between 89°C-91°C, whilst now as you can see in the above image, it reached 93 in no time. I didn't repeat the testing, nor tried to re-enable PowerCut again, to see the result. It was just a quick (and random) observation. You may disregard it for now, until I come with updates.

 

WTF? That TDP puts my skylake chips to shame. Mine barely touches 35W.

 

It's quite strange what happens: overnight the laptop goes into sleep mode and in the morning when I get it on again, PowerCut is disabled, but the CPU will still run beyond TDP - It even reports real power consumption !! (as you could see in the image).
After I restart the laptop, the CPU will go back to its normal locked power mode and won't go beyond TDP unless I re-enable PowerCut. With PowerCut enabled, the power reporting is minimal and not corresponding to reality, of course.
Also, every time I restart the PC, the PowerCut checkbox becomes unchecked...

 

By default, DTS is just an alarm. You can use this feature so when the alarm temperature is reached, ThrottleStop will change profiles, hopefully to a profile with the CPU set to a slower speed so it runs cooler.

90°C or 95°C or even 99°C are all "safe" operating temperatures. Intel CPUs do a fantastic job of staying safe so there is no need to use ThrottleStop to shutdown your computer but if you want to do this, it is certainly possible and quite easy to setup. For my example, I am using a DTS value of 10 which corresponds to a peak core temperature of 90°C (100°C - 10 ). Profile 4 is set up as my shutdown profile. Whenever ThrottleStop switches to Profile 4, the computer will shut down. It will shut down whether the CPU gets within 10 degrees of the Intel thermal throttling temperature (100°C) or it will shutdown if I manually switch ThrottleStop to Profile 4.



Be careful when playing with this feature. If you are using the Task Scheduler to start ThrottleStop when you start Windows and log on and ThrottleStop is set to Profile 4, this is going to force your computer to shut down every time you try to start it. If you ever get stuck in this situation, you would need to boot up into safe mode and delete the ThrottleStop.INI file. In my example, this is not a problem because Windows shuts down ThrottleStop immediately so it does not have a chance to save what profile it just switched to.

That is correct. You do not need to reset VCCIN back to Default. If you do not reset VCCIN, it will just resume from sleep at whatever you had VCCIN previously set to.

86.0 Watts just brought a big smile to my face.
This might not be because of the PowerCut feature. The bios is supposed to set up and restore various power limits when you resume from sleep but sometimes it does not do a great job at this. Some hidden power limits after you resume may no longer be in force. It would be interesting to try this same test without using PowerCut. Set the power limits in ThrottleStop to high numbers, do a suspend / resume cycle and see what happens after you boot up.

I think PowerCut only works when you manually select the PowerCut option. I never got around to writing some code to get the CPU to automatically start in PowerCut mode. It is a powerful feature and I wanted users to be fully aware of what they were doing.

 

@unclewebb Okay, here is what I did:
1.
-set up PowerCut, then I uncheck the checkbox
-sleep/resume: PowerCut disabled, but CPU was still going beyond TDP. Power consumption was being reported accurately.

2.
-set up PowerCut, but left the checkbox checked this time
-again, when resumed from sleep, PowerCut was disabled, but CPU was still overriding TDP like before.

3. Restarting the laptop after a sleep resume will reset the TDP to default values if PowerCut was not re-enabled after it resumed from sleep.

Power Limits are set crazy up in TPL (200W / 300A)

Conclusion:
-the checkbox doesn't stay checked no matter what. Seems to be only a trigger that gets reset. So, having it checked or not, won't matter.
-for some reason, when resuming from sleep, VCCIN is not locked, PowerCut is not enabled anymore, but the CPU is still running on steroids. (by the way, TS reported even more than 90W when running on full stress).

Now, having to enable PowerCut manually every time is a bit of a job, to say so... because I have to close all apps every time, to make sure the CPU goes idle, before enabling it. And this is kind of ruining the purpose of having apps set on start-up. I totally understand it's best for this feature to not be fully automated, especially for "the average Joe", as you like to say (I am not much better than average Joe, but I am trying my best to learn). At the same time, I think this is another situation where command line would come in handy. Calling TS with a parameter that would ensure PowerCut is enabled, would be a good thing and at the same time not too straight-forward like a UI checkbox. So, it would be more likely to be used by an advanced user only.

I am trying to put together as much info as possible, I know having you @unclewebb under fire by so many users can be extremely exhausting and that's why my goal now is to make a tutorial to post it here on the forum about my OC-ing experience. I am not experienced much and when I started this, few months ago, I had no idea where to start from and I was getting sweaty palms only thinking that I may fry the motherboard. But now, as I managed to achieve something, I feel is time to give something back to the community. So, beside trying to get the settings right for my laptop, I want to understand what I am doing, so I can share it further.

 

Now I remember. I think that is the real reason why I decided that ThrottleStop should not enable PowerCut automatically. If the CPU is loaded at startup and power consumption is high, the CPU is going to get locked at a high power consumption level. On a 4700MQ / 4800MQ / 4900 MQ with a 47 Watt TDP limit, this is probably not going to be a problem. When booting up or resuming from stand by, power consumption should never be close to 47 Watts so if the CPU gets locked while reporting 20W or 30W, no big deal. On one of the low power U series CPUs, I thought if power consumption was above the measly 15W or 17W TDP limit, power cut might back fire and lock the CPU while it is already PL1 or PL2 throttling. If this were to happen, throttling would continue indefinitely until the next reboot.

This is a 4th Gen feature and many enthusiasts have moved on to the 6th, 7th and 8th Gen so maybe it is time that I set PowerCut immediately on startup or resume from sleep if the PowerCut box is checked. Would that be a useful feature for you? Just so you understand, all PowerCut is doing is it is locking the VCCIN value. This can only be done if VCCIN is set to a fixed value. If VCCIN is left at Default and you try to Lock it, you get to see a BSOD.

You are not doing too bad for a beginner. You found a way to turn your 47 Watt CPU into a 90+ Watt beast.
Intel CPUs are pretty rugged but some under designed component on your motherboard or your power supply might go pop some day. Keep your eyes open on EBay just in case. I wish I could join the party but my motherboard bios always locks the PL1 / PL2 power limits that ThrottleStop has access to so there are no 90 Watt fun and games for me.

 

Well, as long as there is a way to measure power consumption before activating PowerCut, why don't you use that in your advantage? To time the automatic activation when CPU power is below a certain value (which could be set by the user, depending on their CPU and Machine). Indeed, having some overload at start is highly probable, as Windows will try to load all apps as fast as possible and be ready for use...
Still, the fact that I can have accurate power reporting after resuming from sleep is pretty strange. Do you think that is something specific to my machine/model/bios version, or it could be something more general and which you could eventually exploit?

Thanks, @unclewebb I owe what I learned to your help.

I trust my power supply won't fail, because I am using a oversized unit meant for AW (330W vs 240W which is for M6800), but the motherboard may go off, that I can expect...though I do not wish for it, because it will be crazy to have it replaced (plus there will be windows activation to deal with).

 

Any way to set a voltage offset on Cherry Trail devices? I can set VID if I set the multiplier, but I'd like to undervolt across the whole frequency range.

 

That's strange, I mean I only have two reasons as to why this is not working.

1. The SSD has a driver that disables C states to trick the user into thinking it has lower latency, higher performance etc. Like many people say, manufacturers are lazy and will do whatever it takes to make money. Here its Increase Performance > Better Reviews > MONEYYYYYY!!!

2. Your motherboard/BIOS has a setting that disables C states when the M.2 slot is enabled. My best guess is that the M.2 is taking up an extra PCIe or SATA "Lane", and this causes the CPU to be constantly monitoring the M.2 slot...

Sorry to be pedantic, but have you enabled HIPM+DIPM? Can you tell whether DEVSLP is working on your SDD?

 

Cherry Trail under volting works differently than Core i under volting. ThrottleStop is only going to have very limited support of Cherry Trail because I have no access to any Cherry Trail hardware for testing and development purposes.

No new hardware = no new software features.

@derei - I think your 90 Watt exploit is specific to your machine. I know the trick you discovered does not work on my Lenovo Y510P. Lenovo locked various power limits while your manufacturer left many of the power registers unlocked. What is possible all depends on what a manufacturer has left unlocked. I think sometimes the engineers forget to check things after a suspend / resume cycle or simply don't care.

Timing PowerCut to kick in at just the right time is not something I plan to add to ThrottleStop. PowerCut is one of those cool little features that was mostly ignored by users when it was added to ThrottleStop. PowerCut does not work on any of the recent CPU generations. Intel fixed that bug in their new stuff.

 

@derei If you could send a BIOS dump of your machine it would be great to check the true origin of your unlimited PL.

@unclewebb It would be great if you could perma-enable PowerCut for Haswell users. I am still on Haswell (and Ivy) waiting for the 9th Gen and the 8 core, 16 thread mobile CPUs on 10nm and having PC perma enabled allows me to reach higher clocks.

 

@villahed94 I could provide that, if someone would be willing to investigate, indeed. But as I can see unclewebb seems decided to allow less resources in exploring new options for CPUS that are so old (4'th gen). If anyone will find it interesting to look into the bios dump, I will do that. Unfortunately I am not skilled enough to fully understand BIOS code, so the best I can do is to give it to someone who knows where to look at.

@unclewebb I believe many features in ThrottleStop were not used by many users due to lack of documentation (I still didn't get any response from you in regard to tweaking voltage to Analog I/O, System Agent, Digital I/O - so, basically those are still ignored features by myself, due to lack of documentation). Also, please have in mind that I avoided ThrottleStop for same reasons: was looking intimidating, non-intuitive and lacked documentation. It was only ambition backed up by being disappointed by XTU that made me dig in more... plus your willingness to help. Without your direct support, I would have NEVER used TS.
I don't know how more modern CPUs work in term of PowerCut, but I can tell you there are a lot of people here still using Haswell (is a very robust processor after all) and if adding a timing option is not a (very) complicated feature, it could still be a good idea to add.

There is another aspect that I wish to suggest to you, based on my experience:
As the Voltage Offset is applied equally to the entire curve, there is the well-known issue of freezing or BSOD on low loads due to idle voltage dropping too low. reducing the offset, in order to grant a "safe" idle voltage will cause less performance at high frequencies/loads due to overheating, which may cause throttling (or if PowerCut is enabled, the CPU will go crazy hot).
My suggestion for this is a secondary "Offset" to be applied to the main Offset. How? If the user could set a very narrow range (up to 10mV .. 20mV in total, to say so), that would be applied DYNAMICALLY to the Voltage Offset, depending on load.
For example, If I set a Voltage Offset of -40mV and a the secondary offset to 10mV, then the Voltage Offset would vary between -40mV in full Load and -30mV during Idle. If this feature could be implemented and it would be reactive enough to compensate for voltage, it may actually improve stability at high loads. Basically this would be something similar to "VR Efficiency Mode", to stretch the comparison (yes, I know is not the same, I just tried to oversimplify the comparison), which would make sure there is no voltage drop during idle, whilst the maximum offset is maintained durin maximum load.
Please @unclewebb let me know what you think about this suggestion.

The reason why I came up with the above idea is because I observed I can get very good results with low values (VCCIN: 1.3650 and Core Voltage Offset -41mV) on high loads, where temperature would never go beyond 91-92 Celsius, but the same setting freezes my laptop on low load but when the frequency jumps to maximum for short period. What I had to do in this case was either to increase the VCCIN, or to reduce the Voltage Offset, so it won't cut so much from what VCCIN is providing.

 

Yeah exactly, I expected it to be a manufactor thing, it's a chinese SSD after all.. pretty cheap. However I tested it with my mate's m2 SSD of same kind, but a more recognized manufactorer, and it was the same deal.

I'm guessing the same thing about the BIOS. I don't think there's much I can do about it. I've tried the different link state modes, hipm/dipm etc. I'm not too sure how to check if DEVSLP is working?

 

I feel this is literally impossible to do.
Throttlestop interacts with the Bios. What you're asking for goes FAR beyond bios interaction here.
This is the same as asking a manufacturer to make an adaptive changing loadline slope which dynamically reduces the amount of LLC strength at idle and increases the amount of LLC at load, based on how much power the CPU is drawing, that changes the applied levels (if Bios had LLC1 (low) through LLC5 (high)).

 

If you are referring to my post, then I don't really understand why is so impossible:
1. TS Already sets a certain Voltage Offset. And it can set it "on the fly" ... meaning even with the current profile active and running, one can adjust the voltage offset, press Apply and Voila! - New offset applied.
2. TS can already read the processor frequency/load, etc... so, that data is available
3. Now, what is left, is TS to dynamically manage the Voltage Offset between certain boundaries and apply that setting automatically, without user interaction. Basically to adjust the Voltage Offset slightly based on current multiplier.

So, for example, CPU runs at 40x multiplier and Full Load: this allows me to apply -41mV Voltage Offset. But, if I want the CPU to be stable at 40x multiplier but on Low Load (for example when it jumps from a power saving state to a more efficient one), I would need to set the Voltage Offset to -30mV, in order to prevent occasional crashes. What if this setting would be done automatically by ThrottleStop? Just to make sure the adjustment range is not too big, the user should set a min/max value for this rang (eg. 10mV - this means TS would apply a progressive "taming" to the offset in the range of 10mV from the already set value, depending on load and multiplier).

 

Is chasing after the last bit of 5 to 10mV really necessary?

Enviado de meu Pixel 2 usando Tapatalk

 

This 10mv difference won't make significant difference on temperature while on load... maybe 1c difference at maximum

 

@THEBOSS619 and @Vistar Shook 10mV was in my case the minimum threshold which made the difference between stable and freezing. It was also the difference between temperature steady at 91 celsius, or slowly going to 95 and up... So, it was only one example, not an absolute value. And exactly that is why I suggested a range: if for me 10 mV works, maybe for you 50mV range is what you need.
Point is that if the voltage drops even a bit under the minimum that would grant the CPU to perform normal, it will freeze (or BSOD). So, for who wants to push their machine to the edge, those few mV make a difference.

 

I didn't respond because I never adjust those voltages. Do some Google searching. These voltages seem to be more widely used by people overclocking their desktops. ThrottleStop users need to be way more active in the forums when it comes to helping out with this project. It would be great to see more guides posted like the one Douglas Black did for Notebook Check.

https://www.notebookcheck.net/ThrottleStop-Primer.213140.0.html

@Mr. Fox has also helped users by posting some fantastic ThrottleStop videos to YouTube but where is everyone else? Is it my job to write the software, market it, document it, answer questions on various forums, answer emails and finally; buy new hardware with money out of my own pocket so I can support more hardware? I have decided to take a step back. I do not buy as much new hardware as I used to and I have been ignoring the documentation for years. If users would like to see some better documentation, they need to get together and create some. Any volunteers? I do not mind helping out but I will be spending less time on ThrottleStop this year.

I wish there were more ThrottleStop users just like you. You have taken the time to learn the program and you are still trying to learn what some features do. Overclocking and under volting is for advanced users that are willing to do some digging. It is not intended for users that need to be spoon fed. You learn more by doing some hands on testing. Those are the kind of users that ThrottleStop is for.

I have heard that before but I do not know how to improve this. It is like trying to write a brain surgery book for dummies. Where do you start? Some subjects are too complicated to simplify. That is another problem with having me write the documentation. What makes perfect sense to me will not make any sense to someone that is new to ThrottleStop. Considering the 101 things that ThrottleStop can do and the number of different processor families that ThrottleStop supports, I do not know how to package or document it so it makes sense to a wide range of users. Any suggestions?

As for 2 different offset voltages, I have thought about this but I am not sure if this would work reliably. CPUs change speeds and voltages in hardware in a matter of microseconds. There is no way that software can keep up and manage this. By the time ThrottleStop would notice that the load had changed, it would be too late to react.

 

@unclewebb your TS program is awesome, and it's easy to understand. Remember, I came to your program from back when Coretemp crashed CPU's (I'm sure you remember that), and migrated to a bug free program (realtemp), which the only issues I had were sometimes the EIST and Speedstep settings suddenly not working anymore (and I can't for the life of me remember if this was a bios bug, a system bug or an actual bug in RT), but I personally have always loved your work.

And @Mr. Fox and @Papusan are both one of a kind. There's no one in the world quite like them. I just wish Mr. Fox would get hired at Nvidia and then dethrone Jensen Huang in a power struggle and take over the company. Then we might actually get some REALLY NICE video cards. And Papusan should take over Intel. But I think there are too many BGA printing machines for him to deal with !

 

People should STOP UP thinking... "Should I really support Bad design?"

I put it here... From my last post.


How Dell cripple performance

 

@Papusan while I agree in principle with you, there is also another aspect of the problem: the reality we live in. This are the machines we have, some of us... And honestly, I can't afford now a Clevo or Eurocom machine which seems to have better heat dissipation design (damn, I can't even replace the one I have, why do you think I struggle to get every bit of juice from it? To make it competitive for another 2 years more...).
So, indeed, on long term, there should be a pressure on the manufacturers, but the pressure cannot be made by the users as long as there is no affordable alternative around. You can't simply ask few million people to stop buying laptops until manufacturers will do better... it won't happen.

And yes, as a bottom line, in my laptop (M6800) the heatsink is undersized that it makes me crazy! It was designed SPECIFICALLY for the stock parameters of the CPU. One step further and it can't handle the heat. I can see how the temperature is rising slowly, but steadily, which is clear sign that whilst the heatsink is transferring heat at its maximum capacity, it falls short.
So, this stands to support your message. But I also must say your message is more philosophical than practical.

 

Then expect all you will get is more like this + more of the same expained in my link about crippling above... From worse to even much worse. And it will come the time not even ThrottleStop can save you. Not tomorrow, but it will come be you sure.

 

Um, where are the fans in that thing?

 

Fanless laptops will be more and more normal, bruh. Cheaper to procuce. And easier to clean inside (if you in fact can open the chassis) for the owners. + less parts can break.

 

I7 with passive cooling.....hummm.
Edit: I forgot that intel branded their 4.5W Y processors i7 well...but it seems like the swift will have a throttling 15W U.

 

@derei I can analyze the BIOS and check which Writes to MSRs are being performed or not performed compared to my Haswell machine.
BTW it's really curious to see how cooling got worse on the M6800?
On my M4700 it can handle up to about 65-68W decently and with temps in check.
I have pushed it harder though but reaching cringeworthy temps (in the upper 90s with Conductonaut!)

 

ok quick break from the mini-thin. Can anyone assist with this issue I just noticed. Throttlestop is not monitoring my gpu anymore and I do not know why. The only thing i have done recently is switch from discrete to hybrid and install the new intel drivers. I have since changed back to discrete and this is when I noticed the gpu was not being monitored. help please.

 

When you update Intel drivers, also install/re-install nvidia drivers as well.

 

Well, it may be that you misunderstood a little. Let me clarify it:
-Conductonaut (like yours)
-TS Pushing the CPU to 39X multiplier
-Temperatures going to 90-92 ... and this is just fine as long as it's stable.
But... and there is a twist: This can only happen if the CPU drains less than 85W (TDP is 47W). And I can achieve that ONLY by drastically undervolting it to the limit of BSOD. If I set the voltage to be fully stable for full-range of workloads, the temperature will keep rising even beyond 92 (95-96 in time of minutes).
So, my point is: the stock thermal setup (heatsink + stock thermal paste) would NEVER allow this boost. It wouldn't be capable to drain the heat. Still, even with Conductonaut, the heatpipe seems very much to the limit, not being able to handle it. Hence my statement. (unless you are referring to some other post, and I would need clarification, so I can respond to you in terms).

 

Cool Beans!!! Didn't think to do that. Will go do that right now. Thanks!!

 

This is only a rant, unfortunately... While I can understand your frustration, you didn't provide a realistic model of action. As i said, asking millions of people to stop using their laptops until manufacturers will do better is not realistic. Try it: go to your neighbors and try to convince them of this. See how many would do it. Whilst they may understand your idea, the cost of giving up technology is way bigger than the cost of accepting this crippled tech. How do you think dictatorship works? The same. Fear of loosing their meal for tomorrow is allowing the dictator to steal their freedom.
So, want to solve this? Start producing laptops that are competitive in price for all range of users. There are already (pcspecialist.co.uk, Clevo, Eurocom)... but they are addressing to elite users, whilst the masses are still going to use the mainstream ones. And who is doing most of the sales? The masses.
Anyway, I think we start to deviate the talk from the real situation here, which is ThrottleStop and we should come back on track.

 

Well I tried the driver reinstall and it didn't change the fact that throttlestop refuses to monitor the gpu anymore

 

Well I'm a doof! For some reason under settings the gpu was unchecked. Once I rechecked it POW!!! monitoring mysteriously reappeared. Thanks again for those that tried to assist me with this.

 

I posted a tutorial here which traces and summarize the steps I took in overclocking my Dell Precision M6800, for who is interested to look at it:
http://forum.notebookreview.com/threads/throttlestop-overclocking-dell-precision-m6800.812606/

 

I am always tinkering with my laptop trying to minimize DPC latency on my Dell 9550. Today I did a few tweaks and the DPC latency measured by LatencyMon skyrocketed. Disaster.

Couldn't figure what it was. Then remembered that HWiNFO, once opened, keeps active in the system tray. Closed out HWiNFO and latency returned to normal.

Interestingly, I did not see any negative latency impact from ThrottleStop. Thanks again for a slick program @unclewebb

 

Hey i seem to have ran into an issue everything was great until an hour ago, installed new gpu drivers and after restarting my system throttle stop won't open. just keeps saying can't open winring0.dll, i uninstalled the new gpu drivers which doesn't change anything... it's just i have been dependant on throttle stop for months now due to an issue with my mobo where my clock speeds are locked to 800mhz ... please help

 

Try deleting the .ini file in TS folder and power off, then power on again (don't use the reset command, Power Off!)
Maybe that will fix it

 

i don't seem to have a .ini file in my folder

 

Go to folder / explorer properties and uncheck "hide file extensions for known file types". That's something you've had to do ever since Windows XP.

 

tried that to, still no .ini file showing up

 

Maybe, post a screenshot of what you have in your ThrottleStop folder. Or just delete it, clean up and create a new + tasks. A fresh new start

 

Hi folks. There is a problem with Intel Extreme Tuning Utility. Back then on W7 with my previous notebook this utility was working fine but now I've got a new notebook and it gives me a warning that Intel Watchdog is not present. I've downloaded the utility from Intel's website. Now my question is that what if I ignore this and undervolt via Throttlestop? Will the system revert back to the stable state if things go south? (too much undervolting etc.) I mean are these two programs somehow connected? It is not absolutely necessary but I would be appreciated if I can solve this Watchdog problem too. Thanks in advance.

 

screenshot above, i have tried removing it and downloading again with no luck, what do you mean exactly, when you say, clean up and create new tasks? sorry for the noob questions.

 

just tried TS on my laptop. loads of fine.. something must of went wrong when i installed my new gpu drivers. looking like it's gonna need formatted, can't think of anything else

 

Of course, you could uninstall the GPU drivers and reinstall (or install the previous version. Or, (perish the thought), use a restore point to go back to the state before the drivers were installed - then start over. You should be able to avoid reformatting. Of course, I now have so much software installed, I would do almost anything to avoid reformatting...
Good luck,

joe

 

Indeed, Intel XTU is quite buggy with Intel's own chips lol. Uninstall it.

On most of my systems, I have completely uninstalled XTU and use ThrottleStop, which I have found to be much more stable. Running both programs together can cause conflicts as they try to write to the same registers.

 

Sometimes Microsoft will put WinRing0 on its bad driver list and this will prevent ThrottleStop from opening. A few questions. What is the exact version of Windows you are using? Windows 10 Version 1709 Build 16299.192 seems to be the most recent build of Redstone 3. Are you beta testing Redstone 4? A Windows update might have been installed while you were updating your GPU drivers.

What antivirus program are you using? It might be Windows Defender that is blocking the WinRing0 driver. I use the free version of Avast.

Try moving the ThrottleStop folder to the C:\Program Files (x86) directory.

I have seen this problem before with WinRing0 not opening but I cannot remember what I did to solve it.

 

Hello, I have an Apollo Lake powered Acer Swift 1 (Pentium N4200). On its default configuration, it allows a 6W turbo for 28s, then it throttles down to 4W. With ThrottleStop 8.50's TPL controls (or RWEverything to MSR 0x610) I can raise this to 6W/6W (so 6W sustained forever). But no matter what I try, it will never exceed 6W for any reason or any period of time. Limit resons show PL2 in red. The limits show 0/0, I can set them to large values and they stick but don't work (same for Primary Plane Power limits).

Any ideas on where the 6W might be coming from? It's the nominal TDP for this CPU, but people apparently have been successful getting 15W from it. It might be some other limit from the BIOS (which has no settings as you can expect). It there anything I can try to overcome this limit (using RWEverything or whatever?) Could be integrated into ThrottleStop if it works. Thanks!