The ThrottleStop Guide

Thanks for posting your test results. If your CPU is limited to 56W, then it makes sense that you did not see any change in your CPU temperatures. The cooling system still has to dissipate the same 56W.

There is an obvious increase in your Cinebench R20 scores so using different offset voltages for the core and cache works on the 9300H too.

 

@MrxSiN, what were the laptops scores without undervolting?

 

Average 1600++

 

So. I did a little bit more testing. I notice undervolt does affect maximum CPU frequecny under heavy load. Some said CPU will not achieve its maximum frequency if CPU Core been undervolt too much. Well, i dont think that apply to 9th gen CPU.

System spec, 2019 Nitro 5, i5-9300H
I run the test with C20.

The first test with stock setting.
This is maximum CPU Freq during benchmark when the its trottle down to 45W. I didnt bother with package power before, so I thought it was 56W.

CPU throttle to 3.1Ghz when it throttle down by power limit ending the benchmark with 1562 score.

Second test is Core and Cache undervolt to -150mV

Now, CPU throttle to 3.5Ghz instead of 3.1Ghz. With same 45W package power and ending the benchmark with 1749 score.

Third test with with maximum achieveble undervolt on my system without error in TS Benchmark. -245mV Core and -150mV Cache

Now, CPU Freq is 3.8Ghz while on same 45W.

So, last test, I tried to undervolt everthing else. iGPU, iGPU Unslice -150mV and System Agent -125mV.

Result still the same. CPU throttle down to 3.8Ghz but now I could achieve 4.0Ghz just with 48.4W instead of 50W.

During the testing, time before CPU start throttling down to 45W is varies. Stock setting seem to just hit straight to 45W when start C20 benchmark. On -150mV, CPU stay on 50W for a couple of second then throttle down to 45W until the the test finish. But when undervolt further, its stay on 50W for quite long time before go down to 45W.

 

That statement is not true. Some people in forums do not know what they are talking about. When you undervolt the CPU core, power consumption decreases so the CPU can run faster before it starts to power limit throttle. Your testing proves that undervolting makes your CPU run faster.

What are your turbo power limits set to in the ThrottleStop TPL window? Did you try increasing these limits? Does your CPU still throttle during CInebench to 45W even if the turbo power limits in ThrottleStop are set higher than 45W? Some laptops force a maximum of 45W no matter what power settings you use in ThrottleStop.

 

Hey @unclewebb do you know if BIOS undervolt will keep working after I get the Pludervolt mitigations via BIOS update? Since my MSI BIOS allows me to undervolt via the "hidden" menu that we aren't supposed to access (lol)

 

@Xharos, nobody knows.

 

But there are people who have already received the mitigations, right?

 

Yeah, but you haven't.

 

I set 70 in TPL but it still throttle down to 45W. Look like they force it down.

 

So, I've been playing around with my rig to see what total performance benefits I can discover offsetting the CPU Core Offset from the Cache. I've got some mixed results along with instability that appeared seemingly out of nowhere. I started with the core offset set to -.250 and the cache at -.130. I had excellent performance with an R20 score of 3008 and limited thermal throttling. I was in the process of getting some screen captures to post here when the oddity arose. I set another profile to reflect a -.130 undervolt across the core/cache/iGPU, which yielded a score of around 2600 on R20 and when I selected the profile with the -.250 core undervolt, my system crashed. From that point on, I've been unable to drop the core down below -.150 without instability appearing. I'm not sure what happened and I haven't been able to duplicate that R20 score as a result.

I've been playing with the TPL settings and here are my new, stable and reproducible results:



This score was achieved using the performance fan profile from AWCC (system and CPU fan set to max) with the core at -.150, cache -.130, turbo boost limits clamped at 65w, PP0 current at 100, and PP0 power at 55w (but not clamped). The system did thermal throttle at 100C on two cores and did so multiple times throughout the test.



On this run nearly everything was identical except the turbo boost limits are now clamped at 55w, PP0 current at 80, and PP0 power clamped at 45w. I'm getting an EDP Other flag across the core, GPU, and ring intermittently (which is probably due to the TPL settings I've selected) but during this R20 run, there was no thermal throttling at all. Not coincidentally, the system maxed at 3.7 GHz across all cores. For reference, the 9750H should hit 4 GHz on all cores.

I'm going to keep tweaking this to see if I can get closer to that 3000 score I achieved earlier. I'm also open to any setting adjustments anyone here suggests as I feel like I'm mostly flipping switches like a drunk monkey.

EDIT: According to Cinebench, the 9750H average is around 2600 multicore. Perhaps I should be happy with those results?

 

@Gumwars - Cache undervolts that are too aggressive are limiting 9750H overclocking. Back the cache off to -100 mV so you can see what your core is capable of. Then go back to adjusting the cache. You need to adjust one variable at a time. A 9750H with decent cooling and unlocked power limits is capable of scoring 3000 or more in Cinebench R20 every run.

Never be happy!

 

It's an Alienware M17 - they haven't had decent cooling in almost a decade, lol!

I will continue to tinker then...

 

Agree. If you're happy the retards that manufacture and sell trashbooks win. They do not deserve to win because they are losers.

 

Thanks for your fast reply.

I checked the FIVR window and the EPP value stays as set in throttlestops regardless of the windows battery slider. So that is actually really good to know, because now i have the ease of mind that random processes won't intervene with my cool & quiet strategy. With the windows slider all to the left it sometimes even showed a EPP of 0-63 during IDLE.

I tried to check and log the EPP values during the different windows battery slider positions, but i didn't get consistent enough resultus to post here.

If i just want a cool & quiet system i am fine with an undervolt and EPP setting, right? No need for further throttlestop settings? Not interested in higher clocks or more performance.

 

Hi everyone! I recently bought a Lenovo s540-13iml (i7 10510u 25w and mx250 25w) and i'm trying to get the best out of this laptop. I read a lot about undervolting and optimizing the cpu, but i'm still a newbie, so i'm asking for your help. Here are my settings, is there something i'm doing wrong or that i can do better?
When i run cinebench r20 i get around 1950 and i get those limits in red, so that's why i lowered the turbo ratio.
Also, HUGE thank you to unclewebb, your software is awesome!
https://imgur.com/QoSPxXW

https://imgur.com/CPywxae
https://imgur.com/YfR7VU5
https://imgur.com/NPxmqoO

 

Help Request: Enabling C6 on a 10th gen i3 in a Macbook Air running Windows 10

Hi all (and especially @unclewebb - been using Throttlestop for years and love it).

I have to dual boot MacOS and Windows 10 on a work supplied Macbook Air. They've just supplied me with the 2020 refresh (Ice Lake i3-1000NG4). Windows 10 worked great on the 2019 8th gen CPU, but the battery life is about halved on the 2020 10th gen, and I think it's because C6 isn't enabled for some reason.

I'm trying troubleshooting it over in @Che0063 's battery life/enabling deeper C-states thread - screenshots of Throttlestop, C State window, HWInfo, Task Manager start at this post: screenshots, troubleshooting attempts.

Here's my TS settings + C state info as of now:



Do any TS veterans or people with Ice Lake mobile CPUs have any leads on how I can potentially enable C6? I've received lots of great advice over in the other thread, but figured I'd post it over here in case someone's run into a similar issue before.

Thanks for reading, muchos gracias.

 

@UnspecifiedName - The Intel datasheet shows that your CPU supports package C10.

https://www.intel.ca/content/dam/ww...n-core-families-datasheet-vol-1-datasheet.pdf

There is some C state information in Sections 3.2.4 to 3.2.6 but probably not very useful.

Just some wild guesses but maybe 20% of laptops are using package C8. Maybe 1 in 100 laptops are able to use package C10. Most recent CPUs support package C8, C9 and C10 but you would never know it. OEMs and the people writing the drivers just don't care about the C states.

For my 4th Gen laptop, it is the Intel GPU driver that kills my package C states. I use an external monitor. Sometimes it thinks both monitors are active so it limits the package C states to C2. After I fool around in the Intel Graphics Control Panel, suddenly package C6 starts working properly. That is as good as it gets on this CPU. I have to force it to Single Display.

https://i.imgur.com/vWKYexW.png

It is maddening trying to track down these problems. The one thing you can fix is get rid of useless background tasks. Two core CPUs can spend close to 99% of their idle time in Core C7. Your 66% and 72% numbers in your screenshot will kill battery run time.



The latest version of ThrottleStop should report your BCLK and total MHz correctly. Download, unzip and copy the new ThrottleStop.exe into your ThrottleStop folder if you need this new feature.

https://drive.google.com/open?id=1RYy38wgvtndUhTMM6q5uJsKqt3ZV4NjL

Edit - When Speed Shift is enabled (SST in green), do not check Set Multiplier. It does not work. A Speed Shift EPP value of 80 is better than setting EPP to 180. A slow CPU is inefficient. It will not save power if it has something to do.

 

Thanks so much for your reply, and for all your Throttlestop assistance.

I know what you mean about maddening - despite Ice Lake supposedly supporting down to C10, I still can't even unlock C6. I've tried:

- Selective enabling/disabling of every device in Device Manager, with no luck
- Complete driver update, as many users were reporting that a specific driver was to blame. After updating every single one to the latest and greatest, C6 was still locked.
- Complete Windows reinstall - still no C6.
- Messing with every feature in Advanced Power Settings and Intel Graphics Control Centre.

That still hasn't resulted in C6 becoming available at any point. Out of all the OEMs, I think Apple cares least about C-states - which makes economic sense for them, sure, but it's four hours extra battery life missing on Windows 10 because whoever wrote the drivers didn't care enough to enable them.

Thanks for the tip about Set Multiplier and the 80 value for SST, that's already yielded results in performance.

In regards to the high background task issue for Core C State Percent, I think that's because I took the screenshot when Search was still indexing - after optimisation, I've got the two cores sitting at 96%+ in C7 at idle.

And as far as C State Packages go, it's now about 8% in C2, 86% in C3 and (frustratingly) I just cannot find a way to enable C6. This means that despite all the best possible optimisations, PKG power at idle is 2.2w on Windows 10, compared to 0.3w on MacOS. C0% is down to 0.7, but without being able to drop the package into a lower C-state, that's about as good as it gets.

In the real world, this means a notebook that should be capable of 10 hours on Windows 10 struggles through to reach 6. It's a real pain, but without Throttlestop I wouldn't have even been able to diagnose the issue and optimise what I could, so thanks again for all your hard work and support.

To some extent I understand Apple not optimising this stuff because what do they care, they make their money off their primary OS (even though deeper C-states were working fine for their previous generation notebooks, which is why this is especially frustrating).

But for Windows OEMs to not allow for and optimise modern, deeper package C-states developed by Intel is anti-consumer and wrong.

 

Hi all, my 2020 Razer Blade 15 never goes up to the advertised max single-core boost clock speed of 5.0 GHz for the i7-10750H chip, and I can't figure out why. I've ran multiple single-core Throttlestop Bench Tests and Cinebench R20 tests and none of them make my cores go higher than 4.4-4.5 GHz. I've set the turbo ratio limits to the default factory values of 50, 49, 48, 46, 45, and 43. Also have done some undervolting to the CPU Core (-.250 mV) and Cache (-.105 mV) and the Disable Boost box is unchecked. C-States are already enabled through the registry from the method used in this thread. Gaming Model has been selected in Synapse 3, and Windows power slider set to Best Performance, as well as setting Maximum Processor State to 100%. Is there a setting in Windows that I can tweak to make my CPU run at speeds that Intel said it was capable of?

 

@andrewjazzy - What C0% does ThrottleStop report when your computer is idle at the desktop with only ThrottleStop open? Too much background activity will keep additional cores active. This will prevent your CPU from getting near the maximum multiplier.

Is Speed Shift enabled (SST in green)? What Speed Shift EPP value is reported in the FIVR window? An EPP setting of 0 allows the CPU to run as fast as possible. A higher EPP value will prevent a 6 core CPU from using the highest multiplier.

Post a screenshot of ThrottleStop when idle and post another one when running a 1 Thread TS Bench test. Show me the lowest C0% and the highest multiplier you can get. Also post a screenshot of the C States window when idle.

The ThrottleStop C states window is overdue for an upgrade so it can properly report the 6 core CPUs. An idle CPU should have individual cores spending 99% of their time in core C7.

My Windows install is nice and lean so my idle C0% is quite low. That allows 3 of my 4 cores to remain idle during a 1 Thread TS Bench test. Not a lot of bloatware running in the background on this computer. The 4700MQ can use the 36 multiplier when 1 core is active. The multiplier gets within a hair of this theoretical maximum. At least half of your cores must be doing something in the background.

https://i.imgur.com/O5pQ1AV.png

 

How do you set up your Windows to make it as light as possible?

 

Thanks for the response @unclewebb !

C0% usually hovers around 3-4%, with the lowest value I've seen to be 3.1%. I've disabled all background apps through Windows Settings and all startup apps with the exception of Synapse 3, which is required to be running in order to activate Gaming Mode which increases my CPU max TDP to 60 W (otherwise it caps at 45 W).

Speed Shift is enabled, SST is green, and I use an EPP value of 0.

Here are my screenshots of ThrottleStop at idle and while running a 1-thread TS Bench test:


*The highest multiplier for the 1-thread TS Bench was just slightly above the max value shown here

Here is my C-State window when idle:


Not quite 99% in C7 but I think given Synapse 3 running in the background I can live with it. The question still remains though, how am I still ~400MHz away from the advertised max clock speed? If I can reach something like 4.8 GHz I think I will be satisfied because of Synapse 3, but 4.6 GHz makes me think there is another factor slowing down the CPU.

 

Thank you for the PROCHOT option in 8.75, unclewebb.

On a very thermally constrained Zenbook, I'm setting temp limit to 70 deg C while limiting wattage to 30W and Speed Shift to 0. Trying to get that snappiness at low cpu tasks while not spinning the fan up like a turbocharger, and not spiking the temps all over the place.

 

Just checked c-states on my old Dell 9550 6300HQ. There are 250 tabs open in firefox and a "few" processes running in the background. I usually have EPP a bit higher so c-states are slightly above normal.

All windows updates. BIOS updated through beginning of 2020 but those are blocked now.

 

@UnspecifiedName I get a bit deeper c-states setting window power settings at "balanced" or "power saver". It shouldn't work that way AFAIK.

SpeedShift EPP goes from 0 to 255, for max power and max efficiency. About 80 allows you to hit top turbo speeds but you might find good enough performance and much better power consumption at say 128 or 150 or even higher.

Check out the posts by @Che0063 and @CraftyClown . For example. . .

 

Update: I changed some values, i lowered a bit the uv, because i got a bsod. I also lowered the values in tpl. With the turbo ratio set this way i do not get any thermal throttling, temps at 100% load are 87-88, so i think i can push more, but if i set 37 i triggers prochot. Maybe i can try to set it higher with the last TS beta?

Anyway, sorry if i'm bothering you guys, i'm trying to learn as fast as i can to understand everything, but i'd really appreciate your input, can i do better than this? Am i missing something? Or maybe am i doing something wrong?
Also paging @unclewebb , should i try something else?
Thanks again for your attention, i'll leave my stats here:
https://imgur.com/LHkekdL
https://imgur.com/9wRZDYE
https://imgur.com/fo5eG93
https://imgur.com/MZpIwHv (not getting these red anymore)

 

Hey @pressing - thanks for your tip. I'd actually already come across that post and replicated the steps @Che0063 used, but C6 is still locked. At this point I believe I've tried every combination of drivers/power plans/device manager disables to try and find what's blocking deeper C-states for packages, and I can't figure it out.

I'm thinking it'll probably take a driver update from Apple, and given that Bootcamp drivers from Apple are updated about as often as Satan ice skates to work, there's not much more that can be done at this stage.

 

Greetings,

Firstly, thank you so much to @unclewebb for assisting me a bunch from my last issue. After I did numerous trial and error, I managed to reduce my Alienware m15 R1's temperature a lot and able to increase my battery life.

Alienware m15 R1:
- i7-9750H
- RTX 2060
- 32GB RAM
- 60Whr battery

I undervolted my CPU Core and Cache both to -150mV. On normal office and browsing usage, the CPU temperature stays around 39℃ to 42℃ which is very great! My battery can now last around 5+ hours from 100%. However, I am seeking for guidance with some questions:

1. Can the CPU Core and Cache set to different value of undervolt? Or both must be set to the same value?
2. Is there any other settings I can change to improve my battery life? I append together the FIVR, TPL and C10 windows; hopefully someone can guide me if I can change anything to improve the results.

Here is the link to the screenshots of FIVR, TPL, and C10 windows: https://imgur.com/a/yIRA5dJ

Thank you in advance!

 

Would you mind to elaborate this ?
I set EPP to 76, checking the box, but if I open FIVR I can see Windows is setting 84 instead.
Should I just un-check Speed Shift and let Windows use 84 ?

 

http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/page-1153#post-11016130

Your C state data shows that you have too many apps or one bad app running in the background. Your cores should be spending more time in core C7 when your laptop is idle. This is the easy way to increase battery run time.

Yes. The Speed Shift EPP box is for older laptops or older versions of Windows that were not Speed Shift aware. If you have a newer laptop and Windows can manage EPP then let Windows manage it. There is no need to check Speed Shift EPP in ThrottleStop.

I recently installed Windows 10 -1909 to a 10 year old desktop computer. Its 6 core W3670 was spending less than 0.5% in the C0 state when idle.

Next I used the free O&O Shutup10 tool to disable Windows telemetry and lots of other useless Windows stuff.
https://www.oo-software.com/en/shutup10

Idle C0% is down around 0.2% to 0.3%. No rocket science involved. Windows is not the problem. High idle C0% is always caused by some poorly written app or an app that is running 24/7 in the background. The Task Manager Details tab is the best way to start looking for apps that have gone rogue. The Windows Resource Monitor is also useful.

Some people like to leave Chrome constantly running with 50 tabs open. I prefer keeping Chrome quiet when I am not using it.



Everything has a cost. Users that are interested in maximizing battery run time need to be more aware of how every app on their system impacts C state performance.

 

143W and maximum temperature of 84° ??? How is it cooled ???

Well... 90W is double the rated TDP of those mobile CPUs... it is a reasonable limit IMHO.

 

I just read this on the TechPowerUp forum.

I know everyone loves to upgrade to the latest version of Windows but I cannot understand why. If your laptop is running great, use Sledgehammer to block this update until you know more about what it really installs on your computer.

 

No it’s not.
On Dell gaming laptops is limited to 90W, so you can easily keep 4 Ghz all cores (until thermal limits kicks in).

So basically Windows 10 update 2004 is screwing things up with ThrottleStop???

 

I care about others so I always let them go first. The above was just a warning. If Dell is hiding BIOS updates in Windows 10 20H1, think twice before installing anything. Your laptop might become unusable if voltage control and turbo ratio adjustments are disabled.

 

AFAIK it was the build after this one that TS didn't work in - this one is Slow ring, it was Fast that busted everything. I got a WSL2/Windows Insider Dev to look into it on Twitter.

 

Cold air.

Yes, double the TDP is totally within reason. Around the time of Skylake release ( @unclewebb may know specifically) Intel clarified that the TDP is the maximum NON-TURBO power rating. Actual power usage depends on many factors, including core voltage and core temperature. The higher the voltage the higher the temps go, and as either of those increases so does the wattage.

My 7980XE is 165W TDP (again, non-turbo). But, look at what it does running Cinebench R20 at 5.2GHz (100% overclock). It peaks at about 1282 W (around 0:13 in the video).

 

You are right and I disabled Windows Update (I used Windows Update Blocker, which is just a little bit user friendly than sledgehammer I presume).
But I have a question now: what about security updates ?
Blocking windows update am I preventing any future security update for the life of my notebook? Am I missing something?

 

I have a friend, who has the issue with his Dell XPS 9570 (how could it be different), which was already discussed in the past several times, that his 8750H is stuck at 800MHz and wont clock up anymore. TS shows no throttle indicator, temperatures are all fine and below 50°C. Plugging in the power adapter and it unlocks afterwards. Anyone any idea, what this might trigger, or how to resolve it? Could it be a bad battery giving too less juice and the CPU going into shut down? Would TS show this as throttle?

HwInfo shows a PL2 Power Limit of 90W, I always thought the max would be 78W, as configured in mine. Might this be an indicator, that something is wrong? The configured short and long turbo limit via TS is somehow not shown in HwInfo as Pl1 and Pl2 power limit.

I also noticed that his 8750H has a WAY lower base voltage than mine, how does that come, does every CPU have its own voltage values configured in the CPU itself it uses? How are those values determined? Once in the factory at first test run?

 

PL of 90W is perfectly fine with Dell high-end notebooks.
It is the same on my G7 7790.

 

I am not familiar with that program. Sledgehammer is basically a script that lets me run the Windows Update Mini Tool. With this Tool I can pick and choose what updates to run and when I want to install them. If the tool you are using blocks all updates including the security updates, that would not be a good thing.

Dell usually uses BD PROCHOT throttling but there are 101 ways to throttle a CPU. Seeing some screenshots, including Limit Reasons, allows me to eliminate some of the more common throttling methods.

Using the FIVR - Disable and Lock feature used to screw up HWiNFO. There are multiple sets of turbo power limits. Never sure what set of limits are in control until you see PL1 or PL2 light up in red in Limit Reasons while stress testing.

They seem to.

There should not be a huge difference. Post some pics of your comparison. A lot of the difference might be because of the way your laptops are setup. With Windows 10, you have to be a mad scientist trying to equalize two different laptops for a fair voltage comparison. Slight differences on what is running on one computer compared to the other can vary the VID voltage readings.

 

Like I said, no throttle indicator in TS is shown when this happened, mostly after waking up from hibernate, at some point the CPU was just stuck at 800MHz on battery. No dGPU activity (so no BD prochot), all temps under 50°C, default Windows power slider in middle, speedshift set to 84, turbo limits set to 22/56.

The moment the plugged in the power cord, the stuck at 800MHz went away. I looked up via TV on his machine, and everything was configured normally. Default Windows power slider, all TS settings correct. Intel XTU was never installed on the laptop, not any other tools for power configuration than TS.

Was never endabled/installed. On my laptop, HwInfo also shows 56/78 where I set it to 22/56 in Hwinfo. It is weird, how his shows a PL2 of 90W though in HwInfo where mine was factory always 78W configured.

Both are identically configured, I did it myself, nothing running on both for this obviously, no background services, all properly running. Both proper c8/c10 package power state, both under 1% c0% ect. And under this idle condition, his 8750h runs at ~0.57V where mine around ~0.66. Both use a -100mV offset on core and cache.

I already noticed this the moment I got this laptop, that my 8750H had a way higher, and with that I meant around 100mV higher base voltage, than most other people showing their values in the web and this forum. Mine for example wont ever go lower than 0,66V even with a -100mV offset, where others had values of under 0,5xV under idle condition, with no or normal undervolt offset around -100mV.

So not sure how this is explained, maybe every CPU is tested in factory for a short voltage/power test, and in that test, it is somehow selected, what base voltage values each CPU gets programmed into the CPU, and mine had a very bad silicon so it got a way above average base value.

Here is a screecap he made, sadly with very bad quality, when it happened:


You kinda can see the values from it. VID around 0,57V. All cores stuck at 800MHz. No throttle indicator. Low package power at that moment around 1.5W, some little c0% activity of 2,5%, all temps fine. Speedshift set to 84.

As a comparision my 8750H wont ever go under around 0,66V even with a -100mV and all idle:

 

Ok it just happened again and he called me so I could have a live look on the laptop via teamviewer. The issue is indeed BD Prochot, though all temps are fine, dGPU off. Here are some screepcaps:





What might trigger this? As seen on the picture above, dGPU is off all the time. Broken sensor on the mainboard? I disabled BD prochot in TS and the stuck at 800MHz went away. Connecting the power cord did the same.

 

Unclewebb a question for you.
I tried the above and EPP set at 0, which is good for performance but the CPU was constantly operating at 4/4.2 GHz, with PKG Power of 10/12W and a temperature of 55/60°C. Not a very efficient way to keep a CPU during idle/low duty mode.

I tried SCHEME_MAX and the EPP set at 156. Very power efficient , but relatively poor performance.

I then tried SCHEME_BALANCED, that set EPP to 84 (temperature around 38/40° and PKG Power at 2.5/3W).

Is there another way to use the powercfg /s command line to find a solution in between EPP 0 and EPP 84 ?

 

Download any patches you want as standalone and install in offline mode with Network disabled. Always have the door closed when you install patches.
http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/page-1087#post-10980641

How to use WUB
Windows Privacy Dashboard
Set Network Connection as Metered
NBR Windows 10 Clean Installation Guide
Windows 10 Tweaks and Fixes (Index post #1)
O&O ShutUp10 - Do not use Windows 10 without it!

Edit.
Another option is to use StopUpdates10 if you tolerate the blocking software running in the background.

All roads lead to Rome, Unclewebb. What you use isn't that important as long the goal is exactly the same (Only install what you want - close the door for unwanted changes/ patches)

Sledgehammer is useless without "Windows Update Blocker"

 

That is why I ask people to post screenshots. Pretty obvious when items in Limit Reasons are glowing red, staring you in the face.

BD PROCHOT is just a signal path to the CPU. Different OEMs send different throttling signals down this line. It is not always the dGPU causing the problem. It can be a temperature sensor or it could be a Dell power adapter that has gone bad and is sending rogue throttling signals. It might also be by design. Lots of engineers have come up with throttling ideas like this. The average consumer has no idea how to use a tool like ThrottleStop to figure this stuff out. They just buy a new laptop next year, and next year, and.....

Dell has been throttling their laptops with a wide variety of different methods for over a decade. Up to 90% reductions in performance. Your friend was down 80%. People keep buying them. There is no motivation for Dell to change their business practices. I have not bought a Dell laptop since I started working on ThrottleStop. I think that says it all.

Tell your friend that it is really sad when poor cooling has forced him to run his 8750H at 22W instead of the Intel specified 45W. This same CPU can run at 60W or 70W in a properly designed chassis.

What is wrong with EPP set to 84? This setting works extremely well if you need to lower the CPU speed when lightly loaded.

How is that a fair comparison? His CPU is locked to 800 MHz while your screenshot shows 1037 MHz. You need to lock your laptop to 800 MHz too.

Thanks @Papusan for the info. I talked with a user today who was angry at ThrottleStop. He was running Windows 10 1903 and decided to update to the latest stable version, Windows 10 2004. His Surface Book lost voltage and turbo ratio control. Looks like a firmware update in the process pulled the rug out from under him. I knew those Plundervolt updates would be hitting hard by about now. He wanted me to update ThrottleStop to fix this problem. I had to tell him, sorry but after the update, it is game over.

 

I can’t understand the reason to modify iGPU settings at all. I don’t use the iGPU so I set an offset only on core (-150 mV) and cache (-135 mV).

thank you. Now I’m using Sledgehammer and it relays on Windows Update Blocker.
It is very useful because it allows for a selective installation of patches and a temporary unlocking to use Microsoft Store.

regarding EPP at 84, it is just fine, I’d like to know if there is another method to change the value other than SCHEME_MIN, SCHEME_MAX and SCHEME_BALANCED.
Just curiosity on my side.

regarding Dell, I don’t know which specific model he is referring but I’m having the exact opposite experience on my G7. It is not throttling at all, and basically it is set to run at maximum performance accepting temperatures around 90° C, with both short and long term power limits set by them at 90W.
If I have something to say is that the cooling system could be better to keep temperatures a few degrees lower, but on a relatively thin laptop it’s not an easy task (I don’t understand why every OEM is designing gaming laptops so thin, when they are not supposed to leave the desk very often).
Maybe a different model.

 

I have thinkpad with cpu i5 8265U. it is only a 13''so there is a lot of heat generated and I disabled the driver DPFT after reading about it. The driver remains disabled even after a reboot and the first thing i noticed is that my PROCHOT went up from 90C to 95C, which si great news.

The limit displays PL2 in red while I stress the CPu. SO i would like an advice on how to remove this limit. here is my current configuration

 

According with the screenshot, your PL2 is set at 121W: how is it possible to reach that limit on an i5-8265U ?????

 

@artic_squirrel - Some laptops use throttling schemes that can not be fixed by using ThrottleStop. There is a third set of power limits that ThrottleStop and Intel XTU do not have access to. These limits have likely been set by Lenovo way below what you have set in ThrottleStop. When PL2 is lighting up in red, what does ThrottleStop show for power consumption? Intel's U series processors have a 15W TDP rating. Typical power limits enforced by the OEM might be something like 15W long and 25W short.

You should have bought a Lenovo C930 like I bought for my daughter. No enforced power limits. It is a wonderful device after a ThrottleStop tweak or two.

https://i.imgur.com/eo0J744.png

The latest version of TS allows you to adjust the PROCHOT Offset temperature in the Options window if it was left unlocked. If you want to play with this feature, download, unzip and copy the new ThrottleStop.exe into your ThrottleStop folder and overwrite the previous version.

https://drive.google.com/file/d/1RYy38wgvtndUhTMM6q5uJsKqt3ZV4NjL/view?usp=sharing

If you are letting Windows manage EPP, I think you can edit the Windows Power Profile and tell Windows to use different EPP values. This is a hidden setting. Open up a Windows command window and type in:

powercfg /qh

this will show ( query) all of the power settings for your current Windows power profile including the hidden settings. Have a look in there for Speed Shift related stuff. Windows adjusts Speed Shift from 0 to 100 while the CPU needs to see EPP values from 0 to 255. I think a ThrottleStop EPP setting of 84 is approximately equivalent to a Windows EPP setting of 33.

I know there is a good guide about this stuff. I do not own a laptop that supports Speed Shift so I have not done any hands on testing of controlling EPP using this method.