The ThrottleStop Guide

thank you, great answer as usual.

I tried, but the only related profile I found was SCHEME_BALANCED, the one I'm using, with EPP 84, which is fine.
I ended up using Razer Cortex, a "booster utility" that is basically doing a few things before launching a game (or when I manually activate it): disable some background process, free up RAM (not really needed since I have 32 Gb) and set EPP to 0, just to restore EPP to 84 when exit the game.
Cool and handy.

Basically I'm still playing with it, but your app allowed me to undervolt the CPU, keep temperatures under control (I set an offset in PROCHOT to 87°C, that I feel more comfortable) while keeping constantly good performance during the game, with the CPU running in the 3.7 / 4 GHz range, which is great.
As I said, you should allow PayPal donations in your app, because it is great, and a good job should be rewarded.

 

Happy now? 0,66xV is the lowest my 8750H will ever go, between 800-1200MHz and idle conditions (<1% c0%, <1% pkg pwr), already with a -100mV offset. Where his, and Ive seen lots of other 8750H, all over this forum and youtube, getting to about 0,56V under the same idle conditions with a -100mV offset. Thats 100mV lower. Do I really have to include every little not relevant "information" into argumentations, I know that sentence of yours would come. Trust me, when I say something, that it has values. When I say my CPU has a rough 100mV higher base VID over ALL multiplicators, than that is true. And I wanted to know, why this is the case. I guess no one knows here. So my theory was, that maybe every CPU gets their own voltage VIDs programmed into it, in factory, and mine just has a very bad silicon lottery when it was tested.

Like I said. I dont see ANY throttle reasons of BD prochot being applied on his laptop, NONE. No temperatures too high not even medium, dGPU off, no power cord connected, nothing. So my guess was maybe a bad sensor on the mainboard. It never was ever applied on my 9570.

The reason to do that is to have fans 100% off, not to come around throttling. The 9570 shouldnt throttle at all, especially not to 800MHz, my unit never does it, never. Especially not under idle and cold conditions.

Also all what I said and asked for was not for myself, not my laptop, but for a friends, who cant register at this site, because it seems the site bans IPs from India, for whatever reason.

Also the GUI information in TS is misleading, not sure if this is a bug or intended. But TS doesnt show a throttle indicator with (only) BD prochot being triggered:


So on first look, without the limits GUI open, you dont see there is a "direct" throttle taking place. The throttle radio "button" should always show indicator, when there is any limit happening.

My friend also reported, that the BD prochot option in TS gets randomly reset and activated again, after reboots/hibernate. So this also might be another bug which would be nice to be fixed, that TS always forces a bd prochot. If the laptop activates it again for whatever reason. Save, closing TS, reoping as admin, it is disabled though. But it randomly comes back enabled.

--------------------

I already said it on another post, but wanted to include it here too:

Up from Windows 10 2004, MS removed CsEnabled from registry, you cant revert back to S3 sleep anymore on devices which "support" modern standby. So youre stuck with a broken, not working, never reliable, dangerous "sleep" on your devices forever.

 

I do not have access to your laptop or to your friend's laptop so if you tell me something, it helps if I can see a picture of what you are telling me. A difference in MHz is one explanation for VID differences. When the MHz and load are equalized, one has to conclude that you do have a high VID processor.

Intel's goal is to maximize yield so they can sell as many processors as possible. Going back to the Core 2 Duo days, they have always adjusted the VID voltage curve so they can guarantee that the CPU will run stable at its default specs. I am not sure about what software they use to accomplish this but likely something fairly simple where the CPU is loaded and then the voltage is reduced until the CPU crashes. Add on an extra 100 mV or 200 mV to ensure stability and another CPU is ready for sale. This is done on the assembly line so it has to be a very quick test with lots of room for error. That is why people under volt. If you have the time to do some thorough testing, most Intel CPUs should have no problem running at less voltage compared to what Intel programs into the VID table.

Typically, CPUs with a low VID were considered to be better and more overclockable compared to the same CPU with a high VID. The problem I see with modern CPUs is that power consumption is calculated based on the VID voltage. A high VID CPU might start power limit throttling much sooner compared to a low VID CPU. Maybe you and your friend can run something consistent like 4 or 6 Threads of the TS Bench test at the same frequency and compare reported power consumption. Not sure if Intel does anything at the factory to correct for this possible issue.

That is my guess too.

Just heard from another user today with a Fujitsu laptop. His 10710U just lost CPU voltage control after installing Windows 10 2004.

 

The throttle indicator does not report all types of throttling. The CPU throttle register that ThrottleStop reads does not include BD PROCHOT throttling. I will think about adding BD PROCHOT throttling to this indicator in the future.

I just noticed this bug the other day and I think it is now fixed. Thanks for letting me know. Hopefully the next TS version works more reliably.

 

@unclewebb From the first day I got my XPS 9570, I noticed it had an "above average" VID running over all multiplicators already, as of what I have seen on this forum youtube and reddit by other people, with a 8750H. So my theory was always from the beginning, that I was very very very unlucky, and got a very bad 8750H. I wasnt sure, if it works like this, but it sounds logic, that every CPU is tested in factory, and maybe gets its own voltage curve programmed into it. Some may have an average lowest VID of 0,6, maybe some under average of 0,55, but mine is like on the top above average of 0,77 with no undervolt. I even see sometimes VIDs around 1,3V-1,4V in TS with no undervolt on mine for short period of time.

And my 8750H always got very hot from the first place, going up to 99°C with no undervolt. With undervolt of -100mV it gets to around 92-95°C. And more than -100mV isnt even possible with it, and it crashes easily.

Actually I think there is some hardware damage somewhere in the iGPU silicon too with my 8750H, it happens rarely that the Intel driver crashes with "hardware error" (either LiveKernelEvent 117 or 141) shown under Windows 10 logs, even with no undervolt or very low undervolt of about -60mV, when watching a lot of videos at the same time in chrome for example, or resizing the videos very quickly back and forth for example in VLC.

Back to the 9570 of my friend. He actually told me just before, that the power often goes off where he lives in India, so maybe... that had something to do with it, somehow damaging his laptop, or the Dell power adapter. So maybe it is a bad sensor, bad battery or bad Dell power adapter, which forces the laptop into thinking, "bd prochot" needs to be triggered, for whatever reason.

This garbage XPS 9570 of his was already a mess, I had to fix over the past days, with package power c states all not working for days (all showing 0% in TS), had to flash back several bios versions, do several load bios defaults, to get it back working.

Then modern standby is of course how should it be different a total mes on his unit, creating catastrophic drains, and I cant figure out why so far. Maybe the CPU crashing or it is still c state issue, combined with this bd prochot issue (who knows what happens under modern standby, if the Dell EC reports bd prochot...).

I am on 2004 and undervolt works fine still on the XPS 9570. Maybe it installed a firmware update together with it on his Fujitsu.

 

TS 8.75 beta 5
https://drive.google.com/file/d/1KM4O2iTi6nx3LmxP29ZLe9CodK1oa8G6/view?usp=sharing

- thermal and BD PROCHOT throttling added to the main power limit Throttle indicator.
- fixed BD PROCHOT box so more consistent when starting or resuming.

@maffle - Can you ask your friend to test out this version? The usual install instructions. Download, unzip and copy new ThrottleStop.exe into ones ThrottleStop folder. Thanks for your bug report and feedback.

 

actually at that production volume I don’t think they are testing every single cpu for voltage.
I think the voltage is defined as the one that can feed all the working CPUs exiting the production line. That’s the reason why some (most) are ok even with much lower voltage, while others are accepting just slightly variations, and a few doesn’t accept any variations at all.

did you block firmware upgrade in BIOS security section, on your ?

installing right now.
I’ll let you know.

 

I'm getting 0.55-0.56v at 800mhz when idle.

Other than undervolt, the loadline level the manufacturer uses also affects your voltages post undervolt.

At a low loadlines, the voltage variation between highest and lowest is very small, that means while the maximum voltage is relatively low and the minimum voltage is relatively high. And so you can only apply a small undervolt. What you want is a loadline that has the same headroom at full load and at idle.

On MSI laptops you can fine tune the loadline to make both ends fit exactly to the processor's minimum operating voltage at high and low clocks. I use exactly 1v when the cpu runs at all core 4ghz and 0.55-56v when the processor is idle.

 

@seanwee That is what I already said and observed on several other examples by other peoples laptops with a 8750H or 9750H. My 8750H has a lowest base voltage of about 0.74V though without undervolt and goes on 3.9GHz all core to 1.2-1.3V. Not sure why, but it is how it is. Maybe my 8750H was produced in a different fab and is of lower quality than most others.



This is the lowest mine ever gets, no load, idle. no undervolt. As you see it is dramatically higher, than what I saw on others.

More than -100mV is also not possible with my 8750H and it becomes unstable.

 

My i7-9750H never really stays idle (too many applications like McAfee, Razer and others running in the background, I presume), but it is hanging most of the time in the 900 MHz - 1.4 GHz range, with VID in the 0,58 - 0,6 V range.

 

McAfee is a virus more than an antivirus conidering how much it slows the system. Uninstall it and just stick with Windows defender. They've improved it a lot.

Also clean up your background apps as much as possible. Stop them from starting on boot in task manager.

 

You are probably right about McAfee, but I have an one year subscription already paid, and I will consider to remove it when it expires.
For other services, well, most of them I need, like Razer tools, OneDrive and gaming services like Uplay and Steam.

 

I'd just remove it but I get your buyers remorse so I'll leave it up to you.

As for the other services I switch off autostart and only start them when I need to play a game or change a setting.

 

I just changed the values in TPL window. Of course it is conptlely bogus to put them at 121Watts and so on, but if they are too low, I get PL1 and pl2 limits all the time...

THe power goes up to 23 Watts with PROCHOT at 95C after disabling the in DPFT. With your option to offset the temp, I set 100 deg and the cpu goes up to 26 watts, but the fan is so awful that the laptop shuts off.

I seek advice on how the choice for the next cpu. Is it better to have a cpu rated for 10Watts and then use throttlestop to remove as much limits as possible or get the usual 15Watt cpu and deal with the heat ?

 

All of Intel's low power U series are rated at 15 Watts. How a laptop actually performs with one of these CPUs varies from one extreme to the other. Some have great cooling and some will allow you to use ThrottleStop to set the power limits to whatever you like without the OEM enforcing the 15W TDP limit. Here is how a supposed low power 10th Gen U series CPU can perform in a well designed laptop.

https://i.imgur.com/NQHgK5G.png

As a consumer, I would not want to commit to any laptop unless the store had a really good return policy. That way I could bring it home, do some thorough testing and return it if I was not satisfied. The NBR forums are a great resource. Read up on any new laptop before buying it.

 

Hi, I'm on WIndows 10; 2004 and right after updating I've been getting this error and haven't been able to stop the throttling anymore.
I've checked for programs that are using the RwDrv driver, but I don't think I have any.
I have tried disabling windows defender, but that does not seems to be the issue.

Anyone else on 2004 that is able to use the Disable and Lock turbopower limits ?
I'm on TS 8.75 beta 5

 

The problem is the background apps.


http://forum.notebookreview.com/threads/av-comparatives-performance-test-apr-2020.832837/

 

Perhaps my information is outdated then.

A few months back Mcafee was in the middle or end of the stack and defender was up front.

 

According to that chart, windows defender is much worse.

 

Yeah, at least improved vs before But it's free.
http://forum.notebookreview.com/thr...rformance-test-apr-2020.832837/#post-11011375

 

I heard the new intel cpu will be 9 watts -12 watts, they call it the Y-series

 

thank you, very useful link.
McAfee appears to be very good according to that test. I don't mind paying a few bucks to protect the system (actually 2 of them) as long as performance aren't affected.

Those are low performance / low voltage CPUs mainly intended for ultrabooks. I'm not expecting good performance out of those chips, but it is nice to have 4 cores at around 10W TDP.

according to that article, a few months ago the situation for Windows defender was even worse

 

Windows Defender is still an awful resource hog - one of the first things I turn off.

 

I know about this site rating:
https://www.av-test.org/en/antivirus/home-windows/

Never used anything besides the built in AV in Windows 10 and 8.1, used Comodo a long, long time ago, like 6-7 years ago, until one day it decided to screw with the firewall rules and prevent me to use the internet.
Never again I used third party AVs, never had a virus using just the windows built in AV.

I let it do an initial full system scan after a clean install of the OS and then it just does its quick scans once a week or whatever is the time interval, it takes what 30 seconds, besides that its either at 0.1% CPU usage or not even registering any CPU usage at all.

It will use a lot more CPU time when you are using an HDD, because the scan is slow, and not letting it do ONE initial full scan will leave it permanently tried to do so..

 

It works fine for me. On Win 10 Pro 2004: 19041.264, and TS 8.75 beta 5. The Disable and Lock Turbo box is still checked, and seems normal.
I am using BitDefender for AV. Let me know if you need additional information. - note - this is on my 6700HQ Gigabyte laptop.

 

Thanks for the info. These do not appear to be "new" CPUs. The Y series appear to be recycled cores that were not good enough to be U series cores. If a core is not 100% stable as a U series, easy enough for Intel to lower the CPU speed a little and cut the short term turbo power limit in half to improve its stability. Poof, now we have a low power Y series CPU. If I was shopping for a budget laptop, I would be more interested in finding a U series CPU that can perform like a 45 Watt H series CPU.

The target market for the Y series are netbooks or devices with generally poor cooling. Even if the power limits were left unlocked, it would not be practical to bump a Y series up to 50W. That would be about as much fun for users as trying to hold on to a 50W light bulb with their bare hands.

 

At least up to the 8th gen, even an i7 Y is just a dual core, those are real crap tier dies where almost everything is disabled:
https://ark.intel.com/content/www/u...-8500y-processor-4m-cache-up-to-4-20-ghz.html

Thats an 8th gen die with only DDR3L, cut down iGPU, cut down cache, and only 2 cores, usually you coul find Y series cpu's on very cheap netbooks and ultra portables, or Win based tablets, but with a price of almost 400$ not sure where Intel even unloaded those 8th gen Y..

EDIT:
Found where such things where unloaded:
https://www.notebookcheck.net/Acer-...ht-subnotebook-for-a-fat-wallet.442104.0.html

2k € for 250 CB R15 points..

 

This TS version crashes when you are on AC power and then remove the power and it goes to battery, TS just closes with no error message.

 

Must be something about the way you have it set up. Mine continues to work fine on battery when AC is removed - eve runs TS Bench, albeit with a lower maximum speed.
Best, Joe

 

Are you using the Task Scheduler? Did you follow the guide? Sounds like Windows is crashing ThrottleStop.

http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/#post-6865107

https://i.imgur.com/b4wPCiS.png

 

I am speaking for my friend (-: I will ask him / tell him, if he properly configured the scheduler. Youre right this might mostly be the reason, a bit weird though because it didnt happen before, just after changing from 8.74 to this beta. I think I also found out so far, why bd prochot is triggered for his laptop, or lets say, there is some weird issue with the Nvidia driver or chip itself on his laptop. I was not able so far to figure out, why this happening, but it seems his 1050ti doesnt go into sleep mode during modern standby, and this mostly causes the bd prochot eventually. Or it is the other way around.



I cant imagine this be a hardware issue on the board, this entire 9570 is just a total mess. It was really hard to get the package power c states back working honestly, maybe it is all connected.

 

ThrottleStop never randomly closes by itself. The Task Scheduler option, Stop if the computer switches to battery power, is the #1 reason why this happens.

In the Options window ThrottleStop does have one option that could cause this. The DC Exit Time option. If you checked that box and set it to 1, ThrottleStop would exit one second after switching to battery power. If the Task Scheduler is setup properly, maybe he accidentally clicked that box.

 

A CPU that is not overvolted and kept within Intel electrical specifications is perfectly safe operating close to 100C for an extended period of time.
The problem isn't the Intel CPU. It's the other cost cutting value line components around the CPU and PCB that might have problems with those temps, not the CPU itself.

 

My HP 14-ac001tu Celeron N3050 BD PROCHOT Option is still grayed out on 8.75 beta 5.
https://drive.google.com/file/d/1Zx7PlXum5CYVW_81nO-PLKnuuN76lVR1/view?usp=sharing

 

I have no documentation from Intel and I do not own a Celeton N for testing purposes. That is why that feature is still disabled.

 

Hmm.. I, like @OmarEVG, am also getting the "Error creating service: The specified service already exists" on Win 10 2004: 19041 with TS 8.75 beta 5.

I wonder what would be causing it..

 

In my case it is still working when switch from AC to battery.

I agree. Apparently some manufacturers like Dell, MSI and Apple are confident about their notebooks to operate in that range for extended period of time. BTW almost every gaming laptop is operating above 80º under load.

 

Strange issue on Dell Precision 7740 (last bios 1.9.0) and windows 2004. Since the windows update if throttlestop is launched directly on login I got blue screen with page fault in non paged area. I needed to put on 2 minute delay on scheduled launch.
I already try to pass from TS last stable to TS 8.75 beta 5, same thing.

Edit: 2m delay is OK, 1m still not sufficient. Really odd.

 

Reduce your undervolt, maybe windows is doing something more CPU intensive during boot/login and your undervolt is not stable enough for that.

 

No, I already try also with no undervolt at all, same issue. System once throttlestop is normally open is RS. Seems something wrong in service/driver/bios reading/whatever after login.

 

I'm on a Dell myself (7567, 7300HQ, 1.11 bios), and I also get a bluescreen with that same error if I reboot my computer and then log in with the disable and lock enabled (which is not really working and gives me an error).

All these issues started with 2004

 

Interesting.. yes, maybe can be the "disable and lock turbo power limits" feature (that I also use) that cause the blue screen. Anyway, for me everything is working ok like in 1909 (undervolt, pwr limits unlocks) BUT only with delayed launch.

 

I have also experienced this with newer versions of Windows 10 and Disable & Lock Pwr Limits. I attributed it to Lenovo software, with which it would occur more predictably, but it would still happen every now and the even with the Lenovo software removed. I'll have to try the delayed start.

That said, I can't get the Disable and Lock Power Limits to work at the moment, anyways, because of the "Error to create service" I mentioned above. Haven't tracked down what might be causing that, yet. I'll also mention that I can't read the MSR through RW-Everything anymore (I get an exception), which seems consistent with the Disable and Lock Power Limits failing.

 

@Hopper82 - After you have booted up, can you try cycling the Disable and Lock feature. Clear this check box, close the FIVR window, reopen it and then try enabling that feature. When you press OK, does it show an error message?

I did not write any of the code for this feature or the RwDrv.sys driver that this feature uses. If it is broken in Windows 10 2004, there is nothing I can do about it. If your device depends on the Disable and Lock feature, do not install Windows 10 2004.

 

Ok, I can confirm that the problem is with Disable and Lock feature.

If I launch TS with this feature disabled, no problem even immediatly after login (as usual). Then, if I try to enable 'Disable and Lock power limits' before 2m, again directly blue screen with "page fault in nonpaged area" (no error message in windows before).
For me it's ok also to use the dalayed launch method, anyway.

 

It should be the issue(maybe the features... ) in Windows 10 2004. The problem come from accessing RwDrv.sys or RWEverything. Note that my laptop is HP Spectre x360 13", the 10th Gen, 1065G7 CPU.

About the issue "Error to create service", before I used 8.75 beta 5, that is beta 4, there is no this issue, of course I had ticked the box to use it. After I upgrade to 2004, until I ran beta 5, Windows asked me to not "Run" this program and I chose to run it anyway, this issue come out forever although I ran back beta 4. Thus, I suspected that 2004 may do something good or "memorized" something while the moment to run it "anyway".

About the BSOD, it first come out after upgraded to 2004, after a while, I know that it should be the issue to run RWEverything(I used some script/batch file to set the data in that address, actually it should be the same function of "Disable and Lock feature", I wrote it before that function is available) while startup Windows(both in "Run" setting in Registry and in Task Scheduler), Hibernate back to Windows(through Task Scheduler), or the profile I set in TS to run this script. Also I made a EXE file for this script file for me to run it conveniently everywhere.

However, there is some strange thing. If it is run by just only clicking the EXE file or changing the profile in TS to run it, it is almost BSOD(something like says "PAGE_FAULT_IN_NONPAGED_AREA") every time. However, it will not have problem while startup Windows and Hibernate back to Windows(it only BSOD once so far). I still do not know the reason.

I just aware that if I run RWEverything first, it will not BSOD to run that EXE file ONCE to write data to that address, that is I can see the data is set in RWEverything. However, if I closed the RWEverything and run that EXE file again, it will BSOD again. I also tried once it will BSOD when I just ran RWEverything a time.

Maybe you do not really know what I am talking about, anyway, 2004 will not let you unlock the TDP limitation easily. Good luck everybody.

 

The Disable and Lock Turbo Power Limits feature is the one and only ThrottleStop feature that uses the RwDrv.sys driver.

@cktducky - I know Windows keeps track of bad programs that do not start or run correctly. For some users, after Windows crashes when under volting too aggressively, Windows will put ThrottleStop on the bad list and will prevent it from starting. Sometimes using a slightly different version of ThrottleStop like beta 5 instead of beta 4 will trick Windows into letting ThrottleStop run again. Also moving ThrottleStop to a different directory can also trick Windows into letting it run again.

Thanks for trying to debug this issue. It will probably be a while before I install Windows 10 2004 on any of my computers for testing purposes. I like Windows 10 1909. No issues with this version so no urgent need to install the latest from Microsoft.

 

I’m really happy i followed your suggestion and avoid Windows 2004 upgrade.

 

@unclewebb I've bought the Predator Helios 300 (red version) and it was keeping the Temps about 80~94 ºC. I've put -125 on off-set Voltage settings, and checked the "Disable Turbo" box, but on Cs:Go i'm still getting 80ºC+, in less then 1 hour of gameplay. Is it normal due to my laptop or it's just bad configs that i've maid? Sorry for those dumb questions... I'm new in undervolting computers and i'm kind scared to do something stupid with my laptop. Thanks.