The ThrottleStop Guide

I'm glad I finally got this working. I'm glad to have ThrottleStop running on startup.

 

ThrottleStop: Why not make it 14.95$ then and continue developing it? I'm sure most of the serious users would gladly pay this resonable sum.

 

I noticed the exact same thing! I thought I was crazy lol. I would set my laptops 7820Hk overclock that I knew was stable with a certain - undervolt. Then I would enable speed shift sometimes as I wouldn't always leave it on, And out of no where everytime I enabled speedshift it would "HARD LOCK! I finally isolated the issue around speed shift.

I can only manage a much less undervolt with speed shift. I think its because speedshift already constantly changes voltage, it subtracts the undervolt from the already changing auto voltage from speed shift and it is unstable. idk..

Weird though. Man, I thought I was crazy

W/O speed shift -0.185 undervolt
W/ speed shift -0.090? undervolt

 

I think this is why I cannot achieve a very good under volt with XTU.. I’m getting only about 0.120- before it hard locks.

 

At what speed?

Sent from my SM-G955F using Tapatalk

 

Hey,
thank you for this great app. I managed to get rid of the bd prochot, that set my cpu to 1,6ghz base clock every now and then while gaming.
Apart from that I tried to unlock my Acer Aspire 51g (8250u + mx150) Power Limits without any success. Tried to lock to pl2 limit, which does not make any difference and also tried setting the Powerlimits in rw everything (15900) A0, which sticks in rw app but does not affect real life performance so I guess I am stuck with the bd prochot only.

So, to answer your question regarding that pl lock within your app and the acer aspire 51. It is not locked by default but editing the mmio lines you mentioned and locking it afterwards do not make a difference either even when the line in rw stick. It clocks down to15w after approx. 15 seconds regardless which setting is used.

The friend a few pages back who got his acer swift 3 succesfully unlocked (which got same cpu and gpu) using the rw method makes me guess that maybe he is just using an older bios which is still unlocked. So if I get the bios to downgrade, maybe I will get it to work assuming acer introduced the Pl Limits in some later bios upgrade like lenovo restricted undervolt with some later bios in the mi book pro (I am running the latest one, 1.12 unfortunately)?

Anyone any idea how to trick the acer bios updater to get him downgrading the bios?

any help is gretaly appreciated =)

 

This is why i keep telling people that undervolting isn't the fullproof answer. It's just all most people have access to.
the chips are NOT overvolted.
In fact, they are UNSTABLE if you overclock them at their STOCK DEFAULT VIDS past a few speed bins higher.
People keep saying that "the chips are overvolted from the factory".
No, they are not overvolted. The voltage rises at load because of the reference value for Core IA AC DC Loadline.
there are a bunch of Coreboot (Linux) posts that show what the default value is for kaby lake (should be 2.10 mOhms). Resistance adds voltage. More amps=more current=more resistance=voltage rises.

People have to undervolt because they don't have access to removing this VID boost setting. If you have access to the unlocked Bios or can unlock menus yourself, the solution for all of your woes is changing IA AC DC Loadline to 0.01 mOhms.

Then you get much lower temps, no idle freezes, no freezes when voltage changes, but you may have to overvolt a bit if you are overclocking.
the reason why lower loads/idle voltage causes lockups so often is because the "VID BOOST" is lower at lower current (due to how resistance works). Assuming 80 amps of load with 1.80 mOhms is 80 x 0.0018 (Ohms) = 0.144v VID boost. That's assuming 80 amps of load, so your voltage goes up by 144mv. Pretty close to that -150mv undervolt right?

But at 10 amps of load...that's 0.018mv VID boost only. so now you're reducing voltage by -150mv and only raising it by 0.018v at those very low loads. So you see where problems begin.

 

I have decided to release all future versions of ThrottleStop as freeware with no time limits. That is good news for users but the down side is that there will not be any significant further development of TS. It has had a good run but I am bored with it. Time to move on.

 

oh boy, this spells trouble in the long term for most enthusiastic laptop users here...

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

Maybe someone can pick up where @unclewebb leaves off... or someone takes the reins for a while so unclewebb can get a nice break?

 

Your choice. I'll respect it.

 

I don't think it will work because of NDA. Intel might lock down future CPUs for crippled performance.

 

NDA? huh? with who? unclewebb? intel? im confused here...

 

Time to move on ? Maybe a Throttle stop 2?

 

@unclewebb Completely respect your choice to slow down or cease development of Throttle Stop. I'm a happy user of RealTemp 3.70 on my i7-920 (which I intend to use until win7 sp1 goes out-of-support) and also a happy user of TS 8.5 until my XPS 13 9360 croaks (which it probably will before my i7-920, lol).

I did want to confirm some strange behavior to you though. When my XPS 13 (i7-7500u) goes into hibernate automatically after 5 hours of Standby, and then I wake it from said Hibernation, SpeedShift is disabled and I need to manually re-enable it. Not a huge deal.

 

Goto TPL settings and tick mark enable Speedshift and Use/activate Speedshift when TS starts up.
You need to disable hibernation because it wrecks or causes glitches with TS. Fast startup simply saves your session and goes into deep hibernation.

 

Yes I do have those settings configured as such. However, I do like to use Hibernation as well. Hence the glitch. So far, it hasn't caused any catastrophic issues, so it's not a big deal.

 

Where can i get the 8.60 version of throttlestop? Or it wasn't released for public?

 

Someone else mentioned this problem but I have not been able to duplicate it. I do not have access to a laptop with Windows 10 and a CPU that supports Speed Shift so I cannot do any further testing.

If anyone is interested in trying to troubleshoot this problem, follow along and post your results.

If you are using the Task Scheduler to start ThrottleStop, make sure that you followed the guide in the second post of this thread.

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

If not properly setup, there are some options in there that might interfere with ThrottleStop working correctly. To make sure that the Task Scheduler is not causing this Speed Shift problem, exit ThrottleStop and then start ThrottleStop by opening up the folder it is in and double clicking on the icon to start it.

Open up the TPL window and make sure that both of the Speed Shift options are checked.



Push OK, exit ThrottleStop, restart ThrottleStop and make sure that both of these options are still checked.

With ThrottleStop open on your desktop, make sure that SST is displayed in green.



Try doing a Sleep / Resume cycle and check if SST is still showing green after you resume. Do the exact same test but instead of Sleep, try a Hibernate / Resume cycle. Is SST green when you resume?

In Windows 10 you might have to to go into your Power Options to enable Hibernate mode.

The last test is let the computer enter Sleep mode and then let it go into Hibernate mode. Adjust your power settings so you do not have to wait 4 hours to complete this test. ThrottleStop should still be open on the desktop so when you resume you can see if SST is green or not.

On my 7th Gen desktop computer, Speed Shift is automatically enabled after I resume from Sleep and Speed Shift is also automatically enabled if I use Hibernate mode and resume from Hibernate mode. I do not have a 7th or 8th Gen laptop so I cannot test ThrottleStop when Windows goes from Sleep to Hibernate. In theory if Speed Shift is enabled after both of these tests, it should not matter if Windows goes from Sleep to Hibernate. It should work regardless. Hopefully the above tests can help solve this problem.

Hibernate and Speed Shift work correctly on my 7th Gen desktop Asus board.

It is not quite finished yet. I still enjoy working on ThrottleStop, just not enough time for it lately. The good news is that ThrottleStop should be able to handle all of Intel's recently updated Coffee Lake CPUs like the 6 core Core i9-8950HK, etc.

 

Hello, I require assistance regarding my Zenbook UX331UN Power Limits.

It has a i5-8250U, the issue is that the PL1 Limit is not a constant, but a variable.

From Intel XTU, the CPU has a default wattage of 25W. After a while, the PL1 Limit would slowly start dropping until it is 10W.

This can been seen on both Throttlestop (TDP Throttle) and RWeverything (PL1 Value).

From some of the previous replies here, I learnt that I can use RWeverything to access MMIO memory to remove the Power Limit. I tried removing the power limit, and occasionally it seems like the PL1 limit would remain at where I set it to.

However, most of the times, the value I set for FED159A0 gets reset almost instantly.

My current solution is to repeat setting the values in RWeverything until it stays, but I need a better solution.

 

Try setting the lock bit (Bit[31]) in register FED159A0.

That means when you write a value to this register, the first digit should be an 8. Once that register is locked, it cannot be changed until you reboot or maybe it will be reset if you do a Sleep / Resume cycle.



If this works you can probably write a small script to get RW Everything to do this for you automatically when you boot up. If this does not work, there is probably one more power limit that you do not have access to that is limiting your CPU. Lots of throttling schemes are available to manufacturers.

 

Still trying to figure out how to enable C6 or deeper package C state on MSI laptop. Enable/disable C state in the BIOS changes the Package C State Limit. Enable-C10, disable - C1. I guess MSI lock the C State in hidden BIOS options. I'll get unlocked BIOS after warranty expiration.





Another interesting finding is that according to the review GS73VR with 4K 1060 7700HQ uses 8.4W less at idle compare to the one with 1080P 1060 6700HQ. Maybe MSI enabled C6 in kaby lake?
https://www.notebookcheck.net/MSI-GS73VR-7RF-7700HQ-GTX-1060-4K-Laptop-Review.226124.0.html

 

@Kers - If you ever find where Package C6 is hiding, let me know. It might be a hidden Windows power option that causes this problem. I am curious about one thing. When running on battery power and your computer is idle, does the power consumption data reported by ThrottleStop at the bottom of the screen change any significant amount when you switch from an EPP setting of 255 to an EPP setting of 0? Also keep an eye on the reported Temp data. If the CPU fan speed is consistent, sometimes just watching what happens to the core temperature can give you an indirect indication of CPU power consumption. Less power consumption should show up as lower core temperatures.

I also noticed a couple of things in your screenshot that I wanted to comment on to help others.



The ThrottleStop Power Saver feature depends on the ThrottleStop Set Multiplier feature to lower the multiplier when lightly loaded. Both of these features depend on SpeedStep being enabled.

The next thing that users need to understand is that if Speed Shift is enabled, (SST in green), none of the above old school ThrottleStop features will work anymore. Speed Shift takes precedence over any ThrottleStop feature that is SpeedStep related. If you are using Speed Shift, there is no reason to check the Set Multiplier option. It just wastes CPU cycles.

 

Do what unclewbb suggested awhile ago and check with the file biosbits, to see if windows is blocking the c6 states or something else.
Because even with the fully unlocked MSI Bios, you won't go any deeper than C3.

 

EPP 255 800MHz 2.3W 40 degrees 0.58v
EPP 0 3450MHz 3.0W 40 degrees 1v
MSI does not enable Speed Shift by default and Windows power slider doesn't affect EPP value on my laptop.

I enabled these options cuz I just wanna try random things to see if it triggers C6. Forgot to disable them when taking the screenshot.
Gonna check Windows power option now.

 

Could you please send me a link to unclewebb's post? Not sure what key word to search. Thanks.

 

Powercfg
https://pastebin.com/iabr2SqC

Don't know much about power options but first glance nothing looks wrong imo.

 

Great news, thank you very much !!!

Thanks a lot for clarifying those dependencies and their consequences !

 

https://biosbits.org/

You could have searched this yourself. Not to be rude but I'm not here to do searches for you.

 

@Kers - The CPU power consumption reported by ThrottleStop and all monitoring software is only an estimate and this estimate is not very accurate when a CPU is idle. This data is calculated internally by the CPU and its only purpose is to control the turbo boost function. People put too much faith in this number.

The power consumption number in the ThrottleStop battery box is measured and shows power consumption for the entire laptop when running on battery power. I was curious to see if the measured power consumption changes when going from 800 MHz to 3450 MHz. When idle, cores are spending the vast majority of their time in C7 where they are disconnected from the clock so they are sitting at 0 MHz and they are disconnected from the voltage rail so cores in C7 are at 0 volts. Tweaking Windows 10 and working on improving the percentage of time your CPU cores are spending in C7 when idle might save more power and be more productive compared to getting package C6 working. Even a dual core U CPU can have the 2 cores averaging 99% in C7 when Windows 10 is idle.



Thanks for posting your powercfg settings. I might get lucky and find something hiding in there.

 

@unclewebb
How exactly do you use biosbits to find the available C-states?
I tried it this morning and spent a lot of time, and there were a lot of really confusing, over my head things in there, which seem to be for developers or something.
The only thing I could find was something called "Mwait C1, C3e, c6" or something, and it said "C6 was supported" and it let me "activate Mwait c6, but that's all I could find.

Could find nothing on C10 or anything.

 

I recently purchased an XPS 13 9360, which, unlike the older models, properly uses turbo such that short bursts of power are limited to 51W for 28 seconds and sustained use at ~23W for unlimited time periods is possible and only limited by thermal throttling.

Unfortunately, when using throttlestop, and I tried to change settings, and delete the INI to start over a few times, sustained turbo is for some reason limited to 15W or sometimes 13W and my cinebench scores drop over 100 points vs. XTU. Throttlestop is a nicer program, but as it seems to be messing with turbo functionality, I had to stop using it.

Any ideas?

 

Good question. I must have tried a dozen different versions of the BIOS Implementation Test Suite (BITS) including the most recent version and none of them could report C State residency time on my 4th Gen Haswell. The option was gone. I know BITS used to work so I tried it on a 1st Gen Core i and the Test menu had these options.



The C State Residency option works on older CPUs but the new ones are not supported. Kind of dumb to remove this feature because the newer processors continue to use the exact same registers to store C State data. BITS would be a useful tool if it had been updated but it is dead.

There are power limits in the MMIO (memory mapped IO) that XTU has access to but ThrottleStop does not. That is likely the problem. On your laptop, you will have to continue using Intel XTU. ThrottleStop can solve a lot of problems but not this one. Development of the WinRing0 driver that ThrottleStop uses ended about 10 years ago so with a budget of $0, there is no easy way for me to fix this limitation.

 

Using TS 850, I have enabled it to run on startup and logon by killing the existing login and starting a new one. I have also enabled it to show in tray as i want it to switch profiles between AC and battery modes. However, whenever I login after sleeping the computer, the notification icon disappears but Throttlestop appears to be running in processes. Not sure if it is changing profiles or not, but whenever I resume from sleep I have to restart Throttlestop manually for it to show up again on the nofitications. Any ideas?

 

@unclewebb

There is a random bug in throttlestop 8.50 with my M17x R1, randomly changes the reading of the FSB form 266 to 366 so the clockspeed is wrong unless I hit the FSB button. It randomly happens throughout usage.

 

What does, "killing the existing login", mean? Are you using taskkill to kill Windows or ThrottleStop.exe? I use the Task Scheduler to start ThrottleStop and I have the icon sitting in the system tray. The icon is still sitting in the tray after I resume from stand by or hibernate. The Task Scheduler has a few issues so if you are using it, check out the guide in this thread to make sure that your task is setup properly.

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

ThrottleStop has a few issues like that, especially on the older Core 2 Extreme processors. Being able to unlock and change the maximum multiplier while in Windows was not something that Intel properly documented so bugs like this are inevitable. I have no plans to fix this bug.

 

Sorry meant "killing the existing process" not login... I will recreate the task using your guide and see if that helps. I may have also misidentified the cause of it randomly closing itself as standby because today it happened just randomly. One minute it was in the tray and next minute gone. Hopefully recreating the task will fix it. Thank you.

 

Hi all, do you know how can I set an specific profile as default? It always start with profile 1 but I would like to start with another one.

 

Thank you very much for this app! I'm using it with my i5-6300HQ for almost 1 year. I'm also trying to use it with my i7-7700 (Desktop), but unlike my i5, the speedshift only sets the lower clock limit, so there is no way to limit the clock with Speedshift. For some reason, even being a 65W, it was reaching 86W and 94°C with IBT (no warnings in TPL window). Now, with a -150mv core and cache undervolt, it only reach 63W and 75°C. Another difference is that Package Power Long show as 3408 and PPS as 4095.

https://imgur.com/a/Yp0Bo

 

EPP 0 or 255 result in the same -11.5W ish power consumption. Cores spend 98% time in C7.
Wondering if power consumption number is accurate, what's the update interval? The number won't change even if I start running benchmarks. Exit TS and open again shows the updated power consumption.

 

Yes, doing that will cause problems. When you kill ThrottleStop.exe it will leave a ghost icon in the system tray. This is a Windows feature. After this happens, if you restart ThrottleStop, it might create a new icon and hide it in the "hidden icons" or it might not create an icon at all. Best to not use taskkill on ThrottleStop.

The Task Scheduler has some options in it that are confusing. When not setup correctly, these options can force ThrottleStop to randomly exit. Follow the guide I posted and you should not have any problems.

I do not think that ThrottleStop has a default profile. When your start ThrottleStop, it usually uses the last profile that you were using. It might then switch profiles if you have defined an AC or battery profile in the Options window.

Your motherboard has set the turbo power limits way beyond 65 Watts. This is typical on desktop motherboards. You can set these to the correct values in the bios if you have that option. Package Power Long (PL1) = 65 Watts and Package Power Short (PL2) = 81 Watts are the typical default values for a Core i7-7700. If you open up the ThrottleStop - TPL - Turbo Power Limits windows, make sure it is using these values. If you do not want your CPU going beyond 65 Watts, set both of these limits to 65 Watts and also check the Clamp option. This will force your CPU to a maximum of 65 Watts regardless of what program you are running. IBT or LinX or Prime 95 will all be limited to 65 Watts.

Some motherboards will lock the turbo power limits so you will not be able to adjust them after you boot up. Post some pictures of ThrottleStop if you have any more questions.

Try using BatteryBar.

https://www.techworld.com/download/system-desktop-tools/batterybar-free-366-3213866/

That link should take you to the free version and its power consumption number should update more frequently than ThrottleStop does.

 

Recreated the task following your guide, and it seems to work fine now, thanks!

 

Hello . I just signed up here for help regarding my Intel CPU's TDP throttle situation.

It seems my motherboard's manufacturer (Gigabyte) failed to properly compile the BIOS for it, so now I'm stuck with 65 W PL1. It does not matter what value I choose for the 'Turbo Power Limit' option in the BIOS, the limit is always locked to 65 W. The Intel XTU BIOS works, but I would rather not install that.

Is there anything I can do?

 

@Trendle Can you give us your complete system specs?
What exact CPU? Laptop model? Or desktop model? Did you check the specific Gigabyte forums?

 

@Falkentyne Sure. It's a desktop PC with a Z97-HD3 (latest revision), and I mounted an i5-5675C on it. BIOS version is F10c (I know it's a beta, but I had another issue with the F9 version and the beta version fixed it - no other BIOS's are compatible with my CPU).

I did a search on the Gigabyte forums, but I couldn't find anything.

 

@Trendle Is this an APTIO (AMI) Bios?
Do you have access to something called "CPU VR Settings"->Core I/A Domain?
Is there a setting there called IMON SLOPE?

If this *IS* in fact an AMI Bios, any chance you can make a capsule dump of your bios and upload it somewhere (probably too large to upload here) so I can look at it?

You can dump the bios with "FPTW64 -d biosbackup.bin -bios "without any problems. (run command prompt as administrator to do this).

You can get the file from this mod tools post:
http://forum.notebookreview.com/thr...-clear-cmos-and-prevent-common-issues.812372/

or from my attachment:

 

@Falkentyne It's a 2014 AMI BIOS. It doesn't have any of those settings, as far as I can tell.

When I run FPTW64 I get error 367.

Code:

>FPTW64 -d biosbackup.bin -bios

Intel (R) Flash Programming Tool. Version:  11.0.15.1000
Copyright (c) 2007 - 2016, Intel Corporation. All rights reserved.


Error 367: [FPTw64.exe] cannot be run on the current platform.
Please contact your vendor.

 

Hmm you probably have an older AMI Bios.
This is beyond my ability to help, unfortunately, then. However if you can go through your Bios completely and see if you have a setting called IMON SCALING SUPPORT, and IMON SLOPE (this is the important one), try setting IMON SLOPE to a value of 25 or 50. This will make the CPU report about half the TDP it's actually using and can circumvent the power limit restriction this way.

 

@Falkentyne It doesn't have those settings.

Thanks for your assistance, though. I appreciate it.

 

Is this a Haswell CPU?
http://forum.notebookreview.com/threads/tdp-and-power-limiting-haswell.766743/
Maybe theres something useful here, but all this is beyond me.
Good luck.