The ThrottleStop Guide

I haven't specifically tested it, but it seems every time I reboot (so not often) I need to clean the icon cache and relaunch TS myself.

I have a task to launch it on logon, but it's only in my processes.

Aside from that, I don't really have problems since the update when launching TS normally. (In other words, nothing that hasn't already been solved here)


Envoyé de mon iPhone en utilisant Tapatalk

 

Hey @unclewebb did you do something to throttlestop? I just changed from TS 8.41 to TS Obsidian and I didn't need to run notification area cleaner at all.

I did reboot my PC and not launch TS before I swapped the exes though, but it's still a welcome change. Or is this unintended and I just found the way to bypass it? xD

 

Maybe Windows is confused because it checks only name of the app and not the hashes. Even I pinned TS to start menu, which works flawlessly.

 

I also did that. Usually though it won't get minimized to the system tray after an update but it worked fine this time

Sent from my OnePlus 1 using a coconut

 

The thing is that, TS using the same ini file even after updating it.

 

Yes it would.

Sent from my OnePlus 1 using a coconut

 

I forgot to complete my sentence. @unclewebb already said the upgrading to latest version of TS is super easy, just replace the older executable with a new one w/o worrying about any issues from previous settings stored in ini file. As long as TS reads values from ini files there won't be any problems whatsoever.

 

I know. Usually though one must run notification area cleaner to get it to function properly in the taskbar. Not so with 8.41 to 8.48

Sent from my OnePlus 1 using a coconut

 

I never used Notification icon cleaner ever.

 

Do you make throttlestop sit in your system tray?

Sent from my OnePlus 1 using a coconut

 

I have always had to run the cleanup tool to get it minimized to the system tray... for years. The only people that would not normally need to are those that prefer to have the taskbar icon. I don't want it on my taskbar, only the system tray.

 

hello please i would need an advice.. what do you think it would be a safe value to undervolt my i7 7700? because i started from -120.. and each reboot of my pc logged as well in the windows registry event as ciritical error 41 kernel power.. i would lower it by 5.. i would have no reboots for days and then suddenly something so i kept lowering.. now i'm at -100.. and suddenly i had an other reboot so i lowered it to -90 but i'd rather just lower it to a stable sure value.. of course i had been doing all the tests with all those cpu programs from prime 95 ( even running it for 15 hours) to many more.. but the rig was stable.. evne at -115.. still it seems the only way to see if its stable or not it''s to just use it because i noticed it can happen when using an intense game or even during an operation that does not require heavy cpu usage..
bottom line i would just want to know if anybody has an idea of a safe value to undervolt my i7 7700 without all these tries.. thank you!

edit. not even 1 hour after this post got an other reboot while i was loading a game level.. idk why it's intensifying or it's just a random coincidence.. lowered it to -.80.. i'm getting frustrated :/

 

Exactly! But for some reason I didn't need to this time. I simply didn't run it after a reboot and copied the .exe and when I launched it it worked without me needing to clean

Sent from my OnePlus 1 using a coconut

 

I don't know stable undervolt value for your particular system and chip. Only way to know is to stress test and run daily activities.

But you will see on the Dell 9560 threads that a lot of people with 7700HQ processors had trouble with undervolts in the -130mv range so for those laptops, a less aggressive undervolt seems appropriate.

 

But I don't have the hq.. I have the normal desktop version .. is it the same more or less ? Thank you


Inviato dal mio iPhone utilizzando Tapatalk

 

Sorry about that - no idea.

 

That's good. There is one issue I have been encountering and don't understand why. I have been meaning to post and ask @unclewebb if he has any ideas. This happens randomly, and has been happening across multiple versions of ThrottleStop over the past 3 or 4 months.

Here is the problem @unclewebb...

Upon launch ThrottleStop will sometimes cause my Windows OS to freeze. Happens on W7 and W10. If I delete the INI file (and lose all of my settings) it will start normally. If I rename the INI file, launch ThrottleStop and delete the new one, rename the old one, then toggle ThrottlesStop on/off, change a setting, save the change, close and re-launch things generally return to normal.

Or, another workaround that sometimes work when ThrottleStop decides to cause my system to lock up... if I open XTU so that ThrottleStop is blocked, change some settings, close XTU, kill it, then launch ThrottleStop more than half the time the lock-up problem fixes itself.

This happens completely randomly without any changes to ThrottleStop settings. I don't know if it is a INI corruption issue, a file permissions issue, or what. I initially thought it was maybe a file permissions issue, but it happens on W7 as well as W10, which kind of rules that out since W7 does not suffer from all of the idiotic file permissions problems that W8.X and W10 always have.

Edit: The same issue happens on the Tornado F5 and P870DM3 and those are running different versions of ThrottleStop.

 

I can safely say it never happened to me, though TS 8.41 had some profile setting issues. For example sometimes overclocks wouldn't work, I'd have to tick then un-tick set multi, and for voltages to stick I would have to swap profiles around a bit. I didn't see any of these issues with 8.48 yet so I didn't report them as I only found them a few days ago when fiddling with my CPU.

Do you have the settings in TPL matched to your BIOS? Just curious. The PP0 Current Limit in particular. This probably isn't the issue at all, but I'm just troubleshooting off the top of my head.

 

Hmm... odd that only I have encountered this. It happens to me with the Tornado F5 and P870DM3 and they are running different versions of ThrottleStop.

 

Hey! Both the guide link and download link are dead, any help on this? Got a Y50 and would like to the same.

 

If that's the case then it's definitely likely to be some sort of setting you have in throttlestop, or possibly the BIOS.

Have you disabled the MSR bit lock in the BIOS? I believe mine is set to enabled; it could be causing the random issue (I remember your BIOS set up options indicated you disabled this).

 

How do I contact @unclewebb ? I sent him a "conversation" but is this the same as a private message? I needed to ask him something important!

 

It's completely different then!

Afaik, the chips are quite similar and the desktop chips are allowed to run at higher frequencies due to better cooling in desktop cases.
E.g. the 7700 has got a TDP of 65W and a max frequency of 4.2 GHz compared to the 7700HQ's 45W and 3.8 GHz.
You have to keep in mind that higher frequencies require a higher voltage for the CPU to function to correctly.
Because your 7700 is using a much higher frequency than the 7700HQ, you won't be able to get anywhere near a decent 7700HQ's undervolt when using the maximum frequency.
Should you actually need to undervolt due to thermal constraints or overheating, I'd like to suggest underclocking and using the gained headroom to safely undervolt.

However, I am unable to provde any specific numbers.
Every single chip is a little bit different, so you'll have to try for yourself.
According to this https://www.reddit.com/r/buildapc/comments/5s0p4c/core_i7_7700_has_anyone_tried_undervolting/ thread, some user had decent results when using 1.15V.
You might want to check how much voltage you are currently trying to supply your CPU with

 

could you please explain more? you are suggesting that it's not so recommended for me to undervolt it? i don't have temperatures problem.. i just wanted to lower them a bit.. i'm at -70 now.. hopefully no more bsod and reboots now..
what does it mean decent results using 1.15v? does it mean that they undervolt it with -115?
the thing that i can't understand.. and please if you could explain it.. i've had my cpu stable at -110 with no reboots whatsoever then suddenly i get 2 reboots in 2 days in a row.. i just dont get it.. why is that? coincidence? new drivers updates ? once i find a stable undervolt, shouldn't it be stable for ever?
the voltage i achieve right now.. under full load of all threads is around 1.0250.. that's the value i see in throttlestop..
thanks for your help

 

Options such as Minimize to Tray on Exit is ticked.

 

@Mr. Fox: Did you make sure you're TSC aka TimeStamp Counter clock instead of HPET. Sometimes I observed minor stutters or micro freezing as soon as I click TS.exe pinned to my start menu. Even I have the same issue as you while using XTU.
I use TS as soon as I see Windows Desktop thereby observing some minor stutters, this is my use case. @unclewebb is the related to CPU/OS timers incorrectly being triggered causing unintended issues. One thing I can you tell is, those freezing is similar to HWInfo EC sensor causing audio glitch on AWs. Even I checked using LatencyMon, so far its green most of the time.

 

If the MSR Bit Lock is enabled it will block you from making some changes to overclock settings. MSRs need to be unlocked to have control over them.

It happens on both machines and it is entirely random. May go a month an not happen, then suddenly it does. Then, with a bit of tinkering, the problem goes away again. Very odd.

Yes, I use the built-in timer settings. But, thank you for the info. It could prove useful. When mine freezes up it never recovers. I have to hold down the power button to force it off. Other times it throws a 0x124 BugCheck code (which means CPU voltage is too low) even though none of the voltage settings are changed.

 

Show me a picture of your main throttlestop window.

It is. He is not active right now, clearly, so just wait until he is.

It's really not that different. It's not different at all actually.

The desktop chips are binned better, and don't have piss-poor cooling like laptops generally do. Make the distinction difference: It's not that desktop cooling is "much better" (it definitely can be, though) but that laptop cooling is THAT BAD. Low clockspeeds are par for the course because people keep shoving CPUs into chassis they don't belong in (GPUs too).

The TDP of these chips is a rating, and a limit. It is not how much power it will draw. The 7820HK is a 45W TDP CPU. Set it to 4.7GHz at like 1.3v and run TSBench, suck a cool 80W or something. The difference is whether or not a chip can have its TDP limits unlocked to sustain proper performance. My 7700K can do what it wants. It's up to me to cool it. A 7700HQ may or may not be allowed to pull over 45W (though I don't really think it needs over 45W). A 7820HK should be allowed unlocked power draw, or its potential as an unlocked chip is a waste. Aside from these, the desktop chips are significantly better binned, period. A 7820HK could hit 4.7GHz maybe... at 1.25v+. You could get the same 4.7GHz on about 1.15v on a 7700K, which will produce quite a bit less heat.

The undervolt being worse for mobile chips is due to binning, not due to clockspeed. You should be able to use a 7700 at about 1.05v static, or less.

Ok. Interesting. Will check that.

When your windows freezes, what happens? Does it just lock in place? Sounds almost like what I got when my adaptive undervolt was low and the system was idle, but then... I don't know. I'm testing with static and leaving my PC on with the screen off again to see if it freezes at all. If it doesn't it was the UV, if it does I might have to leave BIOS settings and leave TS off and see if it fixes things.

 

Yes, just locks up or BSOD as I mention in the last post. And, it's random. No change in BIOS settings or ThrottleStop settings. This is why I thought it was a file permissions problem initially. It will come and go without any changes to anything, and stop happening as suddenly as it starts happening. Today is "one of those days" for the Tornado F5 and tomorrow it will probably go back to working normally. Then it will be the P870's turn to act stupid, LOL.

 

Yeah definitely not happening to me like that.

 

Hi there! I need some help Papusan allready gave me useful hints (Thx!), Yet I wonder about my mistake.

Trying to OC ma 6820HK on my Aours (sig) I did the following:

Upped the Short and Long Package Power Limit to 75W.

Added +50mV on Core and Cache.

Then I set the Turbot ratio Limits on all cores as well as teh Cache ratio to 42.

I ran Benchmarks and some prime -> fine.

BUT: In game after some time the CPU drops to base clock, giving me EDP OTHER as LImit Reason for Core and Ring.

Any Ideas?

Best regards

phila

P.S.: Interesting: The System ist oced to 4.0-3.8 Ghz out of the Box throug Aorus Control Center a.k.a XTU. This works fine. Just out of couriosity I started Throttlestop with a clean ini while in game ans soon I had the same drop on CPU clock due to the said limit reasons? Interference between XTU and Throttlestop?!?!

 

Could you explain how to set 1.05 v? I'm just confused about what parameters to change in throttlestop


Inviato dal mio iPhone utilizzando Tapatalk

 

Woah, those are a lot of questions.

Yes, you can undervolt.
Whether the undervolt is needed is a different question.
Usually if something's not broken, don't try to fix it. (unless you enjoy playing with it )

Your CPU is using a default voltage of X for each frequency.
One example (don't fixate on those numbers, they are just an example) is 650 mV for 800 MHz.
In this case, the 650mV, or 0,65V, are X.
If you undervolt, you lower that amount of voltage by your undervolt.
If you undervolt with -100mV, you'll have you CPU supplied by the default 650mV minus your 100mV, resulting in a total voltage of 550mV.
Your CPU is therefore supplied with: X - undervolt = voltage it is actually getting

Throttlestop is not able to set a frequency-specific undervolt - you therefore get the same undervolt for all frequencies.
If you're CPU is clocking at 4.0 GHz and using 1.2V, you are still retracting the same 100mV as you would if it's using 650mV and clocking at 800MHz.
In the mentioned thread, the 1.15V afaik describe the maximum supplied voltage, meaning the CPU clocking at the highest frequency is being supplied with 1.15V.

Your CPU will continuously degenerate over several years and will need an increasing amount of voltage.
So no, your undervolt will never be permanent.
However, the CPU does not degenerate that fast, so I'll have a guess and tell you that your undervolt was not stable in the first place.
You might be able to have a semi-stable undervolt that is fine under certain conditions (Prime) but will crash under the same conditions once the temperature (ambient, GPU, ...) inside your PC changes by a few degrees.
For that reason, it's necessary to test your maximum undervolt and then reduce it by a few notches to have some buffer for changes. (the same principle applies for overclocking)

Have a look at my Throttlestop settings over here: http://imgur.com/a/1j41w

The VID, left part of Throttlestop, is the amount of voltage that your CPU is currently being supplied with. (I've drawn a green rectangle around that value)
The offset, right part, you set up has been retracted from that voltage.

In my example:
VID shows 0.6418V
My offset is -155.3mV
The default voltage (for that frequency, 1059MHz) can therefore be deduced as about 797.1mV.

If you load your CPU, it should jump to its highest frequency.
In my case (second screenshot) it's using 3.6 GHz.
The VID is now showing 1.0448V
My offset is still -155.3mV
The default voltage at that frequency can therefore be deduced as 1.0448V (or 1.05V) + 155.3mV (or 0.15V) = 1.2V

In the same way, you can supply your CPU with 1.05V.
Just load it up, so it uses its maximum frequency and then write down the VID.
Now calculate the offset needed to get your VID down to those desired 1.05V.
(Should you need a formula: 1.05V - VID = offset for undervolting)

But keep in mind: every CPU is stable at a different voltage and you might not be able to supply 1.05V and have a stable PC!
Feel free to play around with the offset and use whichever value suits you best

Enjoy!

PS:

While you might be right about the binning, I have some doubts about the undervolting part.
I have yet to find someone being able to reach even 1.05V on a 7700 while running it at its maximum specified frequency of 4.0 GHz.

Plus: I am not sure whether the higher voltage is needed due to poor binning or simply due to higher temperatures (you are right on the piss-poor cooling compared to desktop builds).

 

Hi Guys! Is there a way to slightly bend DTS or the temp value when Throttling starts?
Under my normal needs for workload temp is around 60-65C where throttling starts and goes as down as 9.00 - 13.00 where in Turbo mode is 22.00

It looks like that when under the load CLICK!
Multiplayer goes down to 12 and temp around 63C... This unit can handle up to 90C. If the temp will be more about 80-85 more heat would be generated and spread over the notebook case and I would be happier

Processor: m3-6Y30 (2.2GHz)
It is about Lenovo Yoga 700 11"
It has terrible cooling properties. Just have a look at this piece of copper CLICK! There can't be done much to make it better.
Machine is pretty nice but with Win10 it's getting terribly slow when heated up after 30 mins using it

 

Have you tried undervolting your CPU?
If not, you might want to give it a try.
By reducing the voltage you supply, you can reduce the heat the CPU releases - and in turn that cute piece of copper doesn't have to carry as much heat as well
Doing a quick search resulted in this interesting post.
So feel free to give it a try.

 

thank you very much for your detailed explanation, i suppose all i can do is just to try.. my latest crash was yesterday at -80, been lowering it at -70 retrying all scenarions from prime95 ( 8 hours straight in blend test) to all my games.. and it's been stable so far.. i'll see in this next week.. hopefully i found my sweet spot.. the temperatures did raise 3-4 degrees more compared to the -100, --100mv i had before but i guess that's ok
I also saw that my cpu is being supplied during prime 95 test with 1,05 volt more or less . that means that all cores were being used 100% and the clock was at around 4 ghz and my cpu ws getting 1,05 volt more or less

 

You need to set core and cache voltage to "static" and set the voltage amount to that, then see if it is stable.
For example, here is mine (set 1.1v):

You are using adaptive. It is likely less stable on 7700s because the lower voltages are reduced too far. Static doesn't drop voltage at lower clocks so the instability is mostly gone.

Binning. Power drawn is increased the higher the temps, but to my knowledge, the temp difference required for voltage needed to change is massive if it exists at all.

 

so do you suggest to me to set those things to static and set my voltage manually like that apart from the values i'm already using to undervolt? what voltage should I set in my case?
could i do any damage to my cpu setting it to static rather than adaptive? thank you!

 

Yeah, if undervolting is not fruitful, try static at a relatively low setting. You can't damage the CPU by setting static, no. It'll simply crash otherwise. As per usual, I do not recommend setting throttlestop to start on windows boot, but rather leaving it to be clicked on. This way you can safely delete the .ini if you screw up the voltages.

For your i7-7700, try 1.05v and see if it's stable. If it isn't stable raise it to 1.06v etc. If it is stable, consider lowering it if you want to try getting better temperatures.

It is easily possible for a CPU using static to downclock and draw very little power, here:

 

thank you! so far i 'll just test this undervolt at -70 and see if it's stable.. my cpu is sucking 1.05v under full load on all threads.. i suppose that is ok?

 

If you're stable sucking 1.05v under any stress test, try using 1v static and see how that works. Your CPU will suck voltage above whatever you set for static, that's how Skylake and Kaby Lake work. So I have 1.1v set, and it sucks about 1.14v-1.15v in stress tests. So consider that offset; try 1v static, see how much power it draws, and see if it's stable. If it is you're good to go. Remember you need to copy static for both core AND cache for it to work in throttlestop. Also, setting a manual voltage does not work with adaptive, for some reason.

 

and what if i have those values mean 1.05 and -70 undervolt with adaptive settings and it's still stable? si that ok?

 

What?

 

I mean, is it still ok if my CPU sucks 1,05 full load and I have adaptive settings enabled and I have -70 undervolt ?


Inviato dal mio iPhone utilizzando Tapatalk

 

Your crashes are likely happening when your processor is idle and your voltage drops.

If you try to set a manual voltage while using "adaptive", it won't do anything. So your undervolt is working off your stock voltage.

You need to set "static" voltage for core and cache, and remove your undervolt, and you'll be fine.

 

all right thank you very much for the explanation i'll try that then

 

I'm new to throttlestop and just noticed the limit reason window. I've cleared it but when coming it shows "thermal" as a reason. I can't find any definition of what this means and the log only shows the cpu getting up to 92c.

Does anyone know what that limit reason is caused by?

 

...
Because your CPU is to hot? I mean, 92 degrees is pretty hot, it would be normal if your PC was set to throttle because of that. Hence, limit reason: thermal.

Just my guess.


Envoyé de mon iPhone en utilisant Tapatalk

 

the CPU isn't supposed to throttle until 97

 

When you have a question, can you post a screenshot that includes ThrottleStop and Limit Reasons and if you have a log file, upload it somewhere like www.pastebin.com so I can have a look.

ThrottleStop can display 3 different types of Thermal warnings in Limit Reasons. Here is the Intel definition when Limit Reasons reports THERMAL.

That isn't a very useful description.

Most Intel CPUs are not supposed to start thermal throttling until a peak core temperature of 100°C but individual manufacturers are allowed to use an offset feature to force Intel CPUs to throttle at lower temperatures. The number beside the ThrottleStop PROCHOT box shows how much offset is being used and a check mark in that box shows if any throttling occurred, even if it was only for a millisecond. Core temperatures can vary rapidly so it is not unusual for the log file to miss the peak temperature but unless the heatsink fell off, the difference is usually only by a degree or two.