The ThrottleStop Guide

The thermal pads conduct the heat from the motherboard. They are the right thickness, so they would make contact with the chassis with or without the graphite sheets, since the graphite sheets are only about 0.1mm thick. The graphite sheets' job is to spread that heat more evenly, as you can see from the thermal images from my previous post.

 

@unclewebb - If i have K series CPU which is 8700K and it allows me to go higher then 95W should i still use "disable and lock turbo power limit" option?

Now i am able to use my CPU at 4,5Ghz with all cores same. TEMPS are 92, TDP also around 90wats with -125volt.

 

Speed Shift is natively enabled in BIOS. Speed Shift Max multiplier is 47 otherwise the OC won't work. The Enable Speed Shift when ThrottleStop starts option is not enabled. Not going to undervolt now because thermal throttling will mess up the stability test. After fixing the thermal issue, I will undervolt using ThrottleStop.

 

@unclewebb - Purely 4.6 Ghz. I wouldn't make it without your help. Thank you reallly much. But with -125 volt temps are around 96. I believe it will throttle at 98C.

 

any plan for macOS?

 

Use this.

 

unclewebb, Throttlestop has a feature to detect usage on battery and being turned off in that scenario but can I ask why there is no feature as "Load Profile # when on battery"?
There is a profile called Battery after all so this feature seems intuitive to me.

 

Some BIOS might be glitchy when querying battery usage that is why its disabled by default. You can check Options and find battery profile when AC power is unplugged and also Load Profile 'X' when on battery.

 

Your results intrigue me! What are your experiences with graphie sheets so far? I got a laptop with uncooled VRMs and was considering using graphite sheets to bridge the heat from VRMs to the heatsink

 

What laptop have you got?

 

XPS 15. Some people used thermal pads to connect VRMs to the heatsinks, but I am interested to see if graphite pads are a better option

 

Ah, they're heatsinking to the bottom cover. It's rather tricky to get all those irregular-height components to make good contact. If you slapped graphite sheet on the VRM you'd touch the inductors but not the MOSFETs. I've used an XPS 15 2-in-1 for a bit and the keyboard deck temperatures could be tolerably a lot higher. So if I had an XPS 15 I would definitely do keyboard-side cooling like I did on the MateBook: cover the keyboard backside with graphite sheets and put thermal pads under the motherboard backside. Oh and if there is any GORE insulation I would remove that. But yeah, graphite sheets can spread quite a bit of heat.

 

They seem to be very flexible, so I could just put it on top of them and keep them pressed in with thermal pads. The main selling point is ridiculously high horizontal heat transfer rate over thermal pads.

I would keep the insulation, but I think having the graphite sheets would be swell

 

Cool, looking forward to seeing what you do with them!

 

I don’t think it will be anything special. Like just a normal repaste results, but with cooler VRMs so it won’t VR throttle. I’ll make a post if I get noteworthy results

 

Only issue with aluminum/heat distribution is that higher distributed heat may mean more failure, as manufacturers tend to cheap out and use other components with lower temperature thresholds than the CPUs are made to withstand.

 

What are some of those components?

 

Stuff like MOSFETS and other components related to power. Im not great in component-level knowledge I'm afraid.

 

Even if you put thermal pad between the CPU and keyboard deck, you'll heat up the latter to at most about 60°C. That's not a significant threat to MOSFETs which can survive over 100°C. What IS a significant and very real threat, is not heatsinking the VRMs to anything. XPS 15, X1 Extreme etc. If you attached their VR components to a 60°C chassis they'd be better off. Right now they're going over 75°C. So it's important to consider that those components are heating themselves more on their own.

Better heat spread, which means lower peak temperature, would actually be healthy.

 

Hi, @unclewebb

Can I ask for the details on the recent ThrottleStop's IccMax control?

My primary OS is Linux and since ThrottleStop is only available on Windows, I need to find a way on how to override IccMax on Linux as well.
Default IccMax on my laptop(LG Gram) is quite low, and hitting all cores even with non-AVX workload quickly throttles down the CPU.
Raising IccMax solves this issue.

I'm very familiar with MSR and MCHBAR, and I already got undervolt and power limit modifications working on Linux,
but I wasn't able to find much information regarding IccMax.

I'd greatly appreciate it if you can help.

Thanks!

 

I am having problems with throttlestop program on i5 8300H, Windows10 x64. Even if undervolt is reset (set to 0), while doing stress test CPU clock speed does not exceed 3GHz. If throttlestop is not working then clock speed during stress test reaches about 3.9GHz. Maybe I have clicked something else by accident? What can I check to fix this and reach full CPU power with undervolt?

I have tried undervolting same values (from -125 to -155) in Intel Extreme Tuning Utility everything works ok, clock speed reaces about 3.8 GHz while stress testing. Maybe I have ticked something unnecessarily in TS. Could someone help me locate that?

 

Maybe.
Post some ThrottleStop screenshots; FIVR, TPL, main screen and lets find out what you have done.

 

I also have the issue when resuming from hibernate my throttlestop config is not in effect. Easiest to tell is from my lowered turbo multipliers. Haven't investigated if my voltage offsets are also not applied after resume. Would be interested to hear updates about that and if you need some tests let me know. I'm running xps 13 9360 with i7 8550u.

 

Here is my setup.
https://imgur.com/a/kfoIWTQ

I have also noticed that there is no screenshot button on my legion Y530..

 

You are concerned about your CPU when stress testing, I ask you to share some pictures, so you show me pictures of your CPU running at full speed when idle. Your CPU is running great in the picture you posted and there is nothing too unusual with your settings.

Try using the FIVR - Disable and Lock Turbo Power Limits option. After that, show me some new pictures when the CPU is fully loaded during a stress test. If the multiplier is not maintaining full speed, include Limit Reasons in your screenshot. Start by using a simple test like the built in TS Bench test.

It will be another week before I have time available to seriously look at this. When Windows goes into or resumes from sleep or hibernate mode, it sends a signal that software can listen to. When ThrottleStop gets a signal that a user has resumed, it normally applies the voltages and the multipliers accordingly. As far as I know, this has always worked correctly until fairly recently. It continues to work correctly on the hardware that I have access to. At the moment it seems like the signal that Windows 10 sends out has changed. It could also be Speed Shift or connected stand by related or who knows.

I do not have an 8th Gen laptop to test with. My plan is to write some code to log what signals Windows sends out during sleep, hibernate and when it resumes. Once that is done, I will make it available so users can do some testing.

 

It seems that after resume the "Turn on" is available in throttlestop. If i click that it still does not apply anything. If i still do turn off and turn on - nothing. But if i open FIVR and click OK there, they apply. I have the lowest radiobutton set for saving the ini.

 

@unclewebb - I was able to hit 5.00Ghz with Cinebench, but with TS bench it gives lots of errors. Does it mean it is not stable ? If i can finish TS test bench means i am stable then, isn't it?

 

I think you need to reset Turbo clocks value in TS. I don't think all cores work at 40x and mostly 1 core clocks at 45x.

 

Some users like the TS Bench test because it does a good job spotting an unstable CPU. Other users do not care if they get a few errors in TS Bench. As long as they are game stable, that is good enough. It is your choice. Give your CPU some more voltage and see if it can pass the TS Bench test. I like being 100% stable which for me includes the TS Bench test.

Other users are reporting the same problem. It will be a few days but I will see what I can do.

 

I feel the same way. All cores fully stable at 4.6Ghz with TS Bench test(No errors at all). If i give more voltage to CPU then temps are getting over 98C and throttling. If i would be able to change this temp limit over 98C would be no problem.

 

@unclewebb hey =)

Been a long time user of TS, and really appreciated the update with disabling the power limits - that helps a lot of laptops CPU to run at full power!

However, I've been experiencing some quirks (on my Elitebook X360 1030 G3) as the service used by the RWDRV file is picky - I now have "Error starting the service: access denied" as soon as TS starts up with the disabling PL feature ON, and had some similar service errors previously. When running TS without this feature, I have no issue.

Would you know which service is it trying to use ?

 

@pau1ow - Are you using the latest version of ThrottleStop from my signature? I know there was a bug with Windows not being able to find the RwDrv.sys file when using the Task Scheduler but this is fixed in TS 8.70.6

 

@unclewebb
Yes, latest version it worked fine for some time but now i got this service error message every time TS starts with the PL feature ON. TS works fine, but my PL are default.

 

@pau1ow - What exact version of Windows are you using and what antivirus program are you using? Are you using the Task Scheduler or just starting ThrottleStop by double clicking on it? Can you post a screenshot of the error message?

I have not done any recent testing but I should have time tomorrow to have a look at some of the recent TS issues further.

 

@unclewebb
Appreciate your support as always
Here we go:
Windows 10 Pro / 1803 / 17134.1
Windows Defender
Starting up TS with task scheduler or manually, both give me the error message

More details below:

 

@pau1ow - I just updated Windows and did some testing.

Windows 10 Pro - Version 1803 - Build 17134.590
Avast Antivirus



Starting up ThrottleStop with the Task Scheduler or starting ThrottleStop manually with the Disable and Lock option checked and both work 100% OK. No complaints.

What is the difference between our systems? Windows Defender.

Some versions of Windows Defender have blocked ThrottleStop before because it uses the WinRing0 driver. Now Windows Defender is blocking the RwDrv.sys because it too can be dangerous.

I do not disagree. ThrottleStop and the Disable and Lock Turbo Power Limits are dangerous features. That is why users love them!

I especially like how Windows Defender pretends that it is Windows that is blocking this driver. It cannot man up and admit that it is Windows Defender that is causing the grief.

I have no plans to change anything to try and make Windows Defender happy.

 

I use defender w/o any issues.
@pau1ow Download configure defender and post a screenie. There are some options that stock WD UI hides. https://github.com/AndyFul/ConfigureDefender

 

unclewebb,

fyi; latest Windows 10 is 1809 Build 17763.316 (as of Tuesday February 11, 2019).

Keep up the great work! Take care.

 

Hahaa.. I'm on Win 1803 as well. v1809 didn't compel me to upgrade. I'm updating it to 590.

 

Dell XPS 9550 - no problems with TS.
​

Windows 10 Pro - Version 1803 - Build 17134.590

Windows Defender Antivirus
TS 8.70.6
​

@pau1ow I had that error from "disable and lock turbo power limits". I needed to carefully reread instructions for additional file that needs to be downloaded and installed. I uninstalled TS, reinstalled, then followed the "disable and lock turbo power limits" instructions very carefully.

 

Hi guys,

So after several W10 reinstalls, I narrowed it down to the FACEIT anticheat (used for CS GO in my case) which, after getting installed, gives me that service error message I was mentioning earlier.
https://www.faceit.com/fr/anti-cheat

If anyone has the patience to install that software, after reboot, I'm positive you will get that error message and thus loose the capability of using the PL unlocking feature

Gonna dig further tonight on what's triggerring the error from this anticheat.

Thanks everyone for the suggestions

 

A guy on the Razer forum was having problems with his 8750H dropping down to 2900 MHz when gaming.

https://www.reddit.com/r/razer/comments/arle3j/throttlestop_rb15_gtx_1070_issue/

It ran Cinebench, etc. at full speed but as soon as the GPU was being used, the CPU would run at a maximum of 2900 MHz instead of 3900 MHz. After lots of testing and some head scratching, he discovered that his Speed Shift EPP setting of 128 was causing the problem. Switching EPP to 80 immediately let his CPU perform as it was supposed to be performing.

The Microsoft default for EPP is 128. If you have Speed Shift enabled, open the FIVR window and have a look in the table at the top right to see what Speed Shift EPP value your CPU is using. I recommend setting EPP to 0 for maximum performance and to 80 when you are setting up a balanced profile. An EPP setting of 128 is not for me.

@pau1ow - Thanks for posting your troubleshooting.

 

I've set ThrottleStop to auto-start with Task Scheduler.
How do I quit it in a clean automated way, right after it has applied undervolt (no need to keep it monitoring in background)?

BTW, I've noticed that with recent NVidia drivers having tools monitoring GPU temps ( HWMonitor, HWiNFO etc.) can cause micro-stuttering.

 

ThrottleStop has an option for that purpose. Go into the ThrottleStop.INI configuration file and add this line.

ExitTime=5

That will force ThrottleStop to exit 5 seconds after it starts up which should give it plenty of time to setup your CPU correctly for you and then quietly exit.

Personally, I think a better option is to just leave ThrottleStop running in the background. The Stop Data option reduces CPU temperature polling to once every 60 seconds which is a microscopic load for a modern CPU. Leaving it running allows ThrottleStop to reset your voltages after you resume from standby.

Did you try monitoring your Nvidia GPU with only ThrottleStop and without HWMonitor or HWiNFO running? Some monitoring apps get GPU temperature info from the driver while other apps access the hardware directly. Just trying to see who is to blame or if this is an Nvidia issue. I know a long, long time ago that some ATi / AMD GPU drivers had micro stutter issues when monitoring GPU temps. Playing with the ThrottleStop - Options - AC Timer Res value can sometimes help with micro stutter issues.

 

Great work and nice option already implemented
Mmmh, actually undervolt is reset after resume from sleep, didn't notice it, thank you. I can always set in Task Scheduler as start condition: On workstation unlock.

I'll investigate further about GPU monitoring tools vs NVidia drivers...
Yes, it was a known issue acknowledged by NVidia with drivers 388… but I won't be surprised if it came back again:
https://forums.geforce.com/default/...h-fps-drops-since-windows-10-creators-update/
https://forums.geforce.com/default/...quick-fix-for-fcu-nvidia-stuttering-issues/2/
https://www.hwinfo.com/forum/Thread...r-MS-Fall-Creator-s-Update?pid=16205#pid16205

 

ThrottleStop does not monitor Nvidia GPU power so maybe it can monitor the Nvidia GPU temps without causing any micro stuttering.

You can leave ThrottleStop running in the background. Turn off the system tray icons and you do not even have to look at it. Its memory footprint and CPU usage on an 8750H is about as significant as a fart in a hurricane.

 

Agreed. Even my mouse configuration software takes up much more resources than TS.

 

Since we're on this topic I have one tiny problem (Omen15, 8750H, GTX1060): Throttle Stop starts by task manager on boot up, all is fine, but when laptop comes back from sleep, sometimes the undervolt is reset, sometimes it is not. When that happens, TS shows in the tray icon, but when I click on it to reset voltage back to low, it disappears and I need to manually restart it to get back my undervolting. It seems to be random and I don't think it depends on the way the sleep was invoked (close lid or time out), but it seems the longer the sleep, bigger the chance of voltage reset and TS unloading. Not a big issue, but makes me wonder why? And why so inconsistent?

 

So in the latest version 8.7.02, it wont save my undervolt voltages, but the last one will.

 

You're not running the latest version.

See:
https://www.techpowerup.com/download/techpowerup-throttlestop/

v8.70.6 is the latest you should be using.