The ThrottleStop Guide

Thanks again, following your advice this seems to work disabled monitoring and "disable turbo" again, and now windows PM seems to be back in control. setting max. cpu at 80% for balanced battery profile now results in max 2.3ghz as before when I started to play around with TS. Plus undervolting both CPU Core and Cache at the same level (-105mV) seems to work. Readings show a clear lower Core Voltage when enabled.
in the main windows .... do "SpeedStep" and "C1E" do have any effect when I have monitoring disabled? Should those be enabled or disabled for best battery performance? Again, maybe it doesn't matter when I have monitoring off anyway ...?
Back to GPU as was mentioned in the next post ... I have the GPU now undervolted -125mV. Will try other tools, maybe GPU-Z to see if I get a separate reading for GPU voltage and install some 3D test tool.

thanks,

boba

 

Hi unclewebb, for me the speed shift is not green like in the mentioned picture, HWinfo also shows it in red. Its a new P651HP6-G Clevo Kaby Lake with Intel i7-7700HQ CPU. I have installed Windows 10 Pro and latest Bios is installed (the one and only available Bios Version at the moment). Is there anything i can do to get it running - before checking with my dealer (Schenker/MYSN)?

 

@DeXa - Thanks for the info.

When Speed Shift is enabled, I do not think that having SpeedStep enabled or disabled matters anymore. I usually leave SpeedStep enabled and C1E disabled. I find that there are so many other C States to choose from that modern CPUs spend virtually zero time in C1E so its status is not that important. Intel recommends that C1E should always be enabled. The Turn On / Turn Off button only controls 3 things. Clock Modulation, Chipset Clock Modulation when the CPU supports that and the Set Multiplier feature. You are not using any of those so everything in ThrottleStop should work OK when you are in Monitoring mode. That button dates back to the early years of ThrottleStop and is still there for backwards compatibility with older processors. For many newer processors, this setting no longer matters.

@AndiiiHD - Some laptop manufacturers have decided not to use Intel Speed Shift technology. I know Dell have decided to leave Speed Shift disabled in some of their laptops. If HWiNFO shows SST in red that means your CPU supports using Speed Shift but the bios has left it disabled. When Speed Shift is enabled by the bios, HWiNFO will show SST in green and ThrottleStop will also show this in green so it is easy to see.

If you have read up on Speed Shift and you want to give it a try, you can use ThrottleStop to enable it. Just click on the ThrottleStop TPL button, go to the Miscellaneous section and check the Speed Shift box and then push the Apply or OK button. Go back to the main screen and you should see SST in green. You will need to reboot to disable Speed Shift. If you decide that you want to enable Speed Shift all the time, open up the TPL window and check off Enable Speed Shift when ThrottleStop starts. After you do that, every time you start ThrottleStop or resume from stand by mode while ThrottleStop is running, ThrottleStop will make sure that Speed Shift is enabled. Pretty simple and hopefully fairly logical once you play with this new setting and enable SST a few times. The SST symbol in ThrottleStop is updated in real time so you will always know the correct Speed Shift status of your CPU with a quick glance. You can also open up the FIVR window and have a look in the monitoring table for the SST status and current EPP setting.

 

Hey again, everything seems fine so far after following your advice. Only thing left is the "start/stop monitoring" setting. Since I want monitoring to be off ... I would need TS to remember that setting, but it doesn't. After a restart monitoring is back to on .... any way to save my settings regarding monitoring?

 

I read somewhere that Intel's GPUs generally couldn't be undervolted as well as the CPU core and cache. is this necessarily true?

I'm only able to undervolt my i7-6500U to -80mV, any value lower and it becomes unstable.
My i5-6200U was about the same, though I could push it to -90mV

My i5-6200's Intel HD Graphics 520 was only able to undervolt to -40mV before crashing
My i7-6500U also has the HD 520 and it can go down to 80mV

What are your experiences on undervolting mobile CPU and GPUs? I know some people have been able to undervolt their CPUs to -160mV

 

The U-series cpu aren't well binned from the start, so you cannot undervolt as much as HQ/HK cpus, let alone desktop K-series.

The intel GPU is very sensitive to undervolting yes, not really sure why.

 

Alright, did some more digging into my problems ...still can't really figure out how TS works in terms of overtaking control over the win10 built in powermanagement regarding cpu frequencies ....after starting TS, it always seems to set CPU max to full speed which is 3.5ghz for me. Though my Win10 balanced profile limits it to 80% (2.3ghz). A workaround I found is to switch to Win10 high performance and then back to balanced which results in win10 taking over again, limiting cpu to 2.3ghz. I found a way doing this through command line with "powercfg" ... so there is a workaround to automate this after starting TS. Voltage adjustment seems to stick now though, which is what I need. will report back as soon as I have this figured out and running the way I need it (just use TS to set my voltages after login and else do nothing).

 

It is easy to misinterpret the speeds shown on the first page of TS - When idling on my 6700HQ, it is common for the speeds to fluctuate in the 3.2-3.5 range for all cores, but if you look under the C7 page, you would see that CPU is spending 80% time in lower states, so those clock speeds are very short term. Another clue is if you look at Pagckage Power on the main page. When idling, the PP rins 1-2 watts, another indication that those max speeds are only transient. Then if you run the TS bench, all of the speeds go up to the max for your CPU (3.1 for mine).
Hope this helps, Joe

 

@boba23 - The Start / Stop monitoring feature was one of my many experimental features. I usually wait for some feedback before I decide if a feature is deemed useful or not. For the next TS release, I will look into saving the state of this button so you do not have to enable it each time.

The original purpose of ThrottleStop was to disable throttling and get the CPU up and running at its full Intel rated speed. ThrottleStop was always intended to be used with the Windows High Performance power profile so I am sure there is some code built into ThrottleStop that doesn't get along very well with the Windows Balanced profile. I will try to do some testing this week to see if I can make ThrottleStop less of a bully when it initially starts.

If you want to limit your maximum CPU MHz when in the Balanced profile, instead of fighting against ThrottleStop why not just let ThrottleStop handle this for you. Easy enough to create a profile with Disable Turbo selected to keep the max MHz down. You could also create a primary ThrottleStop profile with SST EPP set to 0 and Disable Turbo clear for maximum performance and then set up a keyboard shortcut so you can easily switch profiles on the fly or use ThrottleStop to switch profiles based on whether you are plugged in or running on battery power.

As @Jdpurvis mentioned, if you want to reduce power consumption, reported MHz is not important compared to how much time your CPU is spending in the various low power C States. Here is an example of my 4700MQ where the individual cores are spending almost 99% of their idle time in C7.

http://i.imgur.com/iR6ipBR.png

If a core is in C7, it is disconnected from the voltage rail and it is also disconnected from the internal clock so it is actually at 0 MHz and 0 volts. Monitoring software might be showing you 3000 MHz or 2300 MHz or 800 MHz or whatever but that is meaningless if 99% of the time the core is in C7 at 0 MHz. In this situation, the reported MHz might only be accurate 1% of the time so that is not worth worrying about. I tend to ignore MHz when idle. I use the TS - C State data to hunt down useless programs on my computer that interfere with the CPU entering the low power C States. Windows 10 goes rogue sometimes so this is a good way to keep an eye on that too.

A user sent me a pic a while ago of his U CPU and ThrottleStop was reporting that it is using the package C8 C State.

http://i.imgur.com/H19yBZH.png

I am still waiting to see some activity reported in package C9 and C10. I am pretty sure that TS can report these correctly but unfortunately, some manufacturers have disabled the deeper C States and are not using the power saving features that Intel has given them.

 

Hi Uncle,
this one will definitly confirm your assertion: yes C10 is reported by TTS
I7 7500U SPECTRE X360

http://imgur.com/a/ZXSfp

 

@Lordcreator - Thanks for confirming that ThrottleStop can report C8, C9 and C10 residency time. Maybe some laptop review sites should be using ThrottleStop to see which manufacturers are using these low power features.

http://store.hp.com/us/en/mdp/Laptops/spectre-x360-211501--1

The battery run time for those laptops from HP looks impressive.

 

Yes, I can confirm a long battery life, and more importantly, the laptop silence is absolute using undervolting with TTS. The only thing missing is the throttle limit with kaby lake... Maybe for the next iteration of TTS!

 

That would require some laptop review websites to know things about laptops, Unclewebb! You know we can't have that!

 

@unclewebb thanks Speed Shift is now working for me - is it safe to be used without AC connected - i mean when the laptop is running on battery.... i see its using multipliers higher 17 - before it was only 17 or lower without speed shift activated (when on battery)...
and is there any way to enable C9 / C10 states without Premas help (Bios Mods) ?

 

D2 - I just read through your superb "mobile i7 information guide" (linked in your signature 2 posts above). I had a question related to undervolting:

How does one practically optimize undervolting on a fixed current limit laptop? Is that just a trial and error process with random benchmarks, perhaps maximizing stable undervolt, then backing off the undervolt to manually see how temps & power limit throttling "act up"?

For reference, I have a Dell 9550 6300HQ. Ran superhot out of the factory. Better thermal paste, fixed VRAM pads and a -170mv undervolt via throttlestop lets it runs below 70*C in "tough" benchmarks. But it will "powerlimit" throttle at extremes.

 

Yes, pretty much that. You can do some math to figure out a basic deal (let's say you're at 71A fixed limit and at 1v, you can't pull over 71W) but CPUs are a bit... odd and this number won't be perfect.

6300HQ hits TDP throttle? At what power draw? If you can't adjust the current limit and you can't maintain your max power usage, then bumping voltage may very well be all you can do.

 

CPU runs at 45W (max rated) then drops like a brick to ~20W after a few minutes. Undervolt (-170mv) keeps voltages in the 0.9v range. Temp indicators for all sensors are max 70*C or lower

 

Run that at 1v for me and tell me if it pulls more than 20W after a few minutes. -70mV should get it to 1v

 

Thanks D2. I reduced the undervolt from -170mv to -70mv. The throttling mysteriously disappeared in both scenarios; thermal improvement may be due to some tweaks I did this winter (tweaks noted below).

[EDIT - The turbo power max adjustments I recently made in XTU seem to have stuck for my 6300HQ, even after powering off laptop; I see them both in Intel XTU and HWiNFO64. Both set at 65 watts. I also maxxed out both timers to 96 seconds. This could be the reason for no PL throttling.

I'm not sure if ThrottleStop "turbo power max" adjustments are sticking in this but I will check now.

Also before I forget, XTU causes early PL throttling problems with my 6300HQ so I never have it running. XTU also conflicts with ThrottleStop (which I always have running)]

Unfortunately, this does not show if a moderate undervolt can perform better than an aggressive undervolt in a thin laptop. So I would like to find a more aggressive benchmarking tool.

I used Prime95, XTU, ThrottleStop bench, ROG RealBench. Any suggestions to push a bit more? (maybe not FurMark as I am not looking at nuclear option)

_____________
DETAILED COMMENTS

Not sure what caused that performance improvement this winter:

1- The Dell 9550 was a complete disaster at release and the recent Dell-Intel drivers-BIODupdates may have fixed something under the hood

2- I made some tweaks over the past few months (e.g. enabled SpeedShift, enabled "Ultra Performance" (fans-processor boosts) in Dell Command-Power Manager, increased the power limits/times in ThrottleStop...)

ROG RealBench was a stress test that pushed my Dell 9550 to power throttle pretty consistently in a few dozen trials last year. The other tests were relatively low stress. Today I ran the same stress tests at -170mv and -70mv and did not see any thermal or power throttling of CPU or GPU (based on data from HWiNFO64 & GPU-Z).

* CPU max temps were about 68*C and 72*C, respectively
* GPU max temps were about 75*C and 79*C, respectively

RealBench score was about 78,000

==> DO THESE MODS AT YOUR OWN RISK. SEE XTU DISCLAIMER. DO NOT EVEN THINK ABOUT INCREASING WATT LIMITS ON A DELL 9550 WITHOUT ADDRESSING THERMAL PASTE AND VRAM ISSUES.

 

Throttlestop is an "on the fly" adjustment; it does not interfere with your BIOS and only does something when running. XTU writes directly to the BIOS. This is why they conflict. You are right to not use both at the same time.

It is possible your long power limits were skewed in the BIOS and/or your CPU was entering a TDP down state of some kind. I cannot say for sure. But whatever you did, if it is gone now that's great. Note that setting long power to 96 seconds means that stress test programs must full-load your CPU for over 96 seconds to see if it still drops. Prime 95 for about 3 to 5 minutes should do it, or rendering out a long video.

Either way I hope you've fixed it. Skylake mobile should be capable of holding as much power as it wants indefinitely; notebook OEMs simply cripple this somehow after the fact to some degree in general. You might be luckily not crippled to a point where it makes a difference (for example: ASUS locks max TDP to 75W for Skylake chips... this greatly hinders 6820HK and 6920HQ chips, but never 6700HQ chips). Here's to hoping!

Sent from my OnePlus 1 using a coconut

 

Thanks D2. Really appreciate your insight.

My guess is that Dell never intended such huge wattage increases in an i5 running in such a thin laptop. The 9550 thermals were so poorly designed and assembled that this watt mod has a good probability of killing a few laptops. With a good repaste, refitting of VRAM pads, and a good undervolt via ThrottleStop, thermals were quite good in stress tests.

I'm not sure long power will go anywhere near 96 seconds but will give it a test. Regardless...

Finally an error by Dell and Intel in the right direction!

 

No problem, hope it got sorted for you.

 

Time for some Speed Shift testing on a Kaby Lake Core i5-7600. This is a locked CPU so the maximum multiplier is limited based on how many cores are active. When a single core is active, it should be using the 41 multiplier.



I was interested in seeing a comparison between the old school Windows High Performance and Balanced power profiles vs controlling the CPU speed using the new Skylake and Kaby Lake Speed Shift feature. The first test shows how the CPU performs when it is in the Windows 7 High Performance power profile with Speed Shift disabled.



When a single core is active running the TS Bench test, the CPU is able to get as close as possible to the theoretical maximum 41 multiplier. This setting also results in a high reported multiplier when the CPU is lightly loaded or idle.

When switching to the traditional Windows Balanced power profile, the multiplier drops down close to 8 when the CPU is idle but is able to reach the same maximum multiplier when required.



Time to enable Speed Shift. An Energy Performance Preference (EPP) setting of 0 should result in maximum performance and the CPU should behave very similar to the old Windows High Performance profile. When the bios enables Speed Shift, Windows 10 sets EPP to 0 when using the High Performance profile. No surprises here with the CPU once again performing at its maximum speed.



The Intel recommended default setting for EPP is 128. When using the Windows 10 Balanced profile, if Speed Shift is enabled, it sets EPP to 127 or 128. One would think that this should provide CPU performance similar to the old Balanced power profile. Sure it drops the multiplier close to 8 when lightly loaded but look what happens when maximum performance is needed.



Instead of the full 41 multiplier, the new Balanced profile limits the CPU to the 37 multiplier. This is probably why knowledgeable enthusiasts were not impressed when Microsoft decided to enable Speed Shift with an automatic Windows 10 update. Less informed consumers now have a slower computer thanks to Intel and Microsoft. Nice.

The good news is that a knowledgeable ThrottleStop user can adjust the default EPP setting so they can achieve maximum performance similar to the traditional Windows Balanced profile while still being able to take advantage of Speed Shift's ability to get the CPU up to full speed quicker. I decided to drop the default 128 EPP setting down in steps of 16 until I reached 80.

On this CPU, an EPP setting of 80 achieves what I would want to see in a power saving feature. Low multi at idle and a quick transition to maximum performance when needed.

 

thanks a lot @unclewebb maybe this is the reason why its disabled in Bios level on some Gaming Notebooks like my Clevo....

 

@Mobius 1: I am experiencing exactly the same problem you describe on my 6th gen Skylake-U i7 laptop.

See my earlier post and analysis here. Have you managed to solve this in any way?

 

@unclewebb I am in the need of switching to Linux but unfortunately there's no ThrottleStop there. With your MSRFinder tool I think I'll be able to find the voltage offset registers and use some Linux tool to set the offsets by hand. Do you know how offset voltages are encoded in the MSR? I tried to convert from hex to decimal then dividing by 2^13 but no luck; either my math is off or I got the wrong register.

Thanks alot

 

A little update. Two days ago I updated the BIOS (Lenovo T460s - i7-6600U) and the GPU undervolt stopped working. Now I have to change System Agent as well in order for it to work.

 

Got a MSI laptop with a 7700HQ and cant get it to go as high as the claimed turbo speeds of 3800, even if i set the multiplier to 38 T and set it to 38 for 1-4 core active in FIVR, it never goes above 3390mhz regardless of the amount of threads I set it to in the built-in TS bench.

 

can you go to your BIOS and increase the power limits from the stock 200 to 300?

I think there are 2 values, change both

 

@dreassica - If it is maxing out at the 34 multiplier while running the 1 thread TS Bench test then almost guaranteed that you have disabled your C States. These CPUs need at least C3, C6 or C7 to be enabled. On locked CPUs, unused cores need to go into one of these C states so the remaining active core can use the highest multiplier. As always, if you have a question, post a picture or two of ThrottleStop. It makes it a lot easier to troubleshoot.

@todda7 - Voltage adjustment is kind of complicated. I will not be posting or sharing any information about this subject. I don't want to make any more enemies at Intel.

@DeXa - I was playing with a Kaby Lake and for voltage offset to work, the CPU Cache also has to be adjusted. I used to believe that these had to be adjusted equally but it looks like the Cache Voltage can be adjusted to the same value or a greater value compared to the CPU Core voltage. If the core is set to -100 mV then the Cache voltage has to be at least -100 mV too but it can be undervolted further to -150 mV or whatever. CPU Cache seems to work independently so I can set only that and this will work fine. I will have to play with the Intel GPU to see how that one works. I haven't quite figured out the exact relationship between all of the individual voltages. The bios update likely contained a new microcode that fixed a bug or two within your CPU. The FIVR window shows what microcode version the CPU is using. If you had any old screenshots you could compare.

 

I never touched the c-states, and they are for sure enabled in the BIOS.
Got it to go to 3600 with 1/2 thread now, still not 3800 though.

 

Set your SST=0

Uncle Web had a post on that earlier this week

 

So I locked these a while ago.. can I unlock them again*?



Anything specific I should change in the ThrottleStop.ini file?

 

@dreassica - Listen to @pressing. You have Speed Shift enabled. For maximum performance go to the main screen and check off Speed Shift - EPP and set that to 0. For more of a Balanced profile, instead of 128, try using EPP = 80. The default setting of 128 can limit maximum performance. This post on the previous page explains what is going on.

http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/page-548#post-10450191

When Speed Shift is enabled, there is no need to check the Set Multiplier option. It is also unlikely that your laptop uses Clock Modulation throttling so do not check the Clock Modulation box unless a Log File specifically shows that your laptop has this throttling problem. Checking these boxes just wastes CPU cycles fighting throttling methods that are not being used. The Kaby Lake desktop CPU I am playing with can be under volted at least -100 mV for the CPU Core and CPU cache. Definitely look into using ThrottleStop to do some undervolting if you are interested in maximizing performance, decreasing power consumption while on battery and reducing your laptop's maximum temperatures. Win, win, win!

@jaug1337 - It is kind of a hassle when you select Lock but change your mind later. You will need to exit ThrottleStop, delete the ThrottleStop.INI config file, completely shut down your laptop, start it back up, run ThrottleStop and think twice before selecting the Lock options next time. I should put a better warning on those options since once the CPU is locked it is a hassle getting the CPU unlocked.

 

You are a god on earth... a thousand thanks to you!

 

I have the new XPS 13 2-in-1 with the Kaby Lake Y series i7. . I'm able to go to -1000mv on the core.. Seems to do nothing there, but the cache undervolting DOES work and will crash after more than -85mv.

ThrottleStop 8.40 being used, and it shows it's -1000mv undervolted on the core on the right side and that's been applied..

Is this a matter of the BIOS blocking undervolting? Makes me sad because they are preventing the ability to run the chip with more aggressive voltage offset :/

 

any idea why Limit Reasons is greyed out for me after a clean installation of Windows + Throttestop 840 ?

 

@AndiiiHD

It would be helpful if you could show some screenshots/information, including what your CPU is.

 

Oh yes - I forgot to mention. Its a Clevo Laptop http://www.clevo.com.tw/clevo_prodetail.asp?id=984&lang=en P651HP6-G

 

Dunno, maybe it is blocked by the BIOS? I don't know whether the limit reasons is detected by the BIOS or software

 

@Diversion - I found on my Kaby Lake desktop CPU that the core voltage can be adjusted but it cannot be adjusted beyond what the CPU cache is adjusted to. In your case, the CPU cache is at -85 mV so you can only adjust the CPU core to -85 mV and that is it. The CPU will ignore any core voltage adjustment request beyond that. That is why you can set it to -1000.0 mV. Anything beyond -85 mV is being ignored. Your GPU setting is also being ignored. I think on your CPU you have to adjust the Intel GPU and System Agent equally. Wish I had more documentation and hardware to play with to confirm this but I don't.

@AndiiiHD - Limit Reasons works on Kaby Lake CPUs but I forgot to unlock the Limit Reasons button on the main screen. My mistake. That's what happens when I am programming without access to the appropriate hardware for testing purposes. I depend on user feedback but no one mentioned this problem. I have to constantly surf the net looking for tiny scraps of info because users continue to ask questions without posting any screenshots. I ended up joining a few Chinese and German forums but I don't know a word of either.

Anyway, this problem was fixed a while ago. I should have time to release an updated version in a few days. I got frustrated waiting for some meaningful feedback so I bought another CPU that I really didn't need to buy. More money down the drain. Maybe my dog will share his bowl with me when I am old.

 

@unclewebb

Limit Reasons working:


Seems like on battery, the new XPS 2in1 turns on BD Proc Hot to keep it running cool/save battery.

 

It appears System Agent + IGP do something and right now i'm not certain if it's just undervolting the system agent AND IGP or just system agent by itself but I verified by crashing the system at -125mv on both IGP and -125mv on SA. These are my current stable settings:


Funnily enough; undervolting the Y series doesn't seem to really help anything at all unfortunately. It likes to begin throttling after the first 3 seconds no matter if the TDP is low or not.. And I can't get it to hold any frequency nor lock the multiplier to anything, it just bouncing up and down like crazy all the time. I think it's just the nature of a CPU that has a TDP of 4.5 watts but can boost to 12 watts on TDP-up. The only time I can get the CPU to hold to a speed is by disabling Turbo and it will hold 1.3ghz the entire time with no bouncing. Turbo bins are another ordeal...

 

@Diversion - Did you try clearing the BD PROCHOT box in ThrottleStop to disable the throttling? If you do this while stress testing on battery power, you might permanently damage your battery so proceed with caution. Without being one of the engineers that designed your device, it is impossible to say what they were trying to protect.

I think you have been spoiled by your 6700K. The non K CPUs constantly adjust the amount of turbo boost so it is not unusual for the multi to wander endlessly. It is too bad that the low power CPUs could not be tweaked some more. The built in power limits are very restrictive and difficult to get beyond.

Can you post a screenshot of TS? Not sure if these CPUs support Speed Shift.

 

It's no big deal really, as in normal usage the temps remain in high 30s mostly.. Just trying to eek out the most battery life I can via undervolting.. Seems in addition to the rest, I was also able to go to -80mv on the Analog agent. I'm actually pretty impressed with the Y series on a whole, being 4.5watts and beating the i5-7200U I had in my previous XPS in single threads and getting pretty close in multi-core tests as well.

Y series isn't going to be for anyone who needs long-term CPU performance, such as rendering/encoding, this thing seems to settle around 2500~mhz after long periods of loads (which is a full 1000mhz slower than it's max speed).

 

http://ark.intel.com/products/95441/Intel-Core-i7-7Y75-Processor-4M-Cache-up-to-3_60-GHz

Intel lists your CPU as 4.5W with a TDP Up of 7W and a TDP down of 3.5W so it sounds like you are doing OK. I am sure your undervolting will get you some more battery run time.

These CPUs are not for everyone but they seem to be a big step up from the previous Atom CPUs that were used in the low power devices. The Atoms were more like reworked Core 2 technology from a decade ago.

 

Dell has some sort of magic going on, I can hit an upwards up 12watts using the ThrottleStop 32M test. Probably explains why my scores are so close to the i5-7200U which is a 15 watt CPU. Dell said that when plugged in it allows more wattage and more speed as well, so I guess this is what i'm experiencing.. Although far less of it on battery (which i'm fine with).

 

@unclewebb thank you for your amazing software. With its help I was able to reduce temps significantly and also prevent my 6700hq from throttling.

I am wondering if there's any way to donate to you?

 

Donate is a dirty word on NBR but I do appreciate the offer. I also appreciate hearing from happy users. This project isn't making me rich so some positive feedback keeps me going. Hopefully the next version will be ready in a few days.

@Diversion - What does the TPL window show? Maybe Dell has tweaked the TDP Level limits in the bios.

Edit - I found an interesting quote on the [H]ard Forum.

I am not sure why it works but sounds like he's happy!

 

@unclewebb