The ThrottleStop Guide

Hmm interesting - I had read that results improve after 24-48 hours, but I found that not to be true with my Gelid Extreme re-paste, and after 6 days I redid it with Phobya liquid metal. I don't think I noticed any improvement when I re-tested right before re-pasting though, so maybe it does differ with different pastes? Or maybe different laptops simply have different pressure levels due to their range of tolerances. Mine seemed to have good contact because when re-pasting again, there was very little paste left in between the sink and the parts, a few ripples but most had gotten pushed out, whereas on the factory job, there was loads of paste in between.

 

Sorry for the extra post - I am evaluating a replacement unit Dell provided, not sure if it's worth delving into. From the factory, the paste job must be much better, because this one can hold 4.2Ghz with the undervolt of -125, no problem. Once I get up to 4.3 or 4.4, I start getting Power Limit 2. At first this seemed weird, total wattage wasn't even at 85 watts, but then I realized a core further down the list was briefly touching 92 (had to scroll to see it). Whereas on my first unit, one core was constantly above 90, if not a couple of other cores also, before I re-pasted.

Here is my question - even if the new unit doesn't undervolt more (or not much more) than the old one before blue-screening, is it still possible I will get more speed out of it? On my old one, with liquid metal, max temp using CB 15 repeatedly is 83 on one core, all others are 78 and below. But I cannot go higher because 4.4Ghz starts hitting the 110W power limit built into the chip. And I can't reliably undervolt more than -125mV to use less wattage.

But on this new laptop, with similar undervolt, power limit doesn't seem to get anywhere near 95 watts, let alone 110, even at 4.4 Ghz (before the thermal limit kicks in). Is this even possible? Or is it because I hit the thermal limit first that the wattage doesn't get to the full usage (i.e., the power limit of 110w takes a while to build and kick in, so I am not seeing it yet, although it's about to be tripped if the thermal didn't kick in first)?

Put another way, if I do all the work of re-pasting and perhaps re-padding, is it expected to be the same as my first unit with the same undervolt - it will have trouble going above 4.4 Ghz? Or is it possible for two CPUs to have different max stable overclock boosts on the SAME undervolt (and same for all other settings)? I suppose what I am wondering, is this -- if the fact the voltage is set to adaptive, means that one CPU might run more efficiently and use less voltage in its adaptive setting, regardless of what the undervolt is set to, and thus achieve higher clocks than another specimen that doesn't run as efficiently. Is this possible? Or are the individual specimens' efficiencies only reflected in how far they can be undervolted (and in that case, same max undervolt means same max power/same max clock)?

 

my temps are stable 1 hour playing game, 3,4ghz frequenchy (temps 80,81,82)c but sometimes it can reach to 89c for 5second & then down to 80-82C

 

I7 7700HQ ?

 

nope, i5 8250U bro

 

Dear Uncle, I'm using TS on my shiny new NUC8i7BEH with an i7-8559u CPU that runs fast and hot.
I'd like to use the profiles to either limit power consumption to a minimum or go full speed ahead when I choose, but I'm running into an obstacle.

Speedshift - EPP does not seem to 'listen' to the profile setting. When switched on it lowers the CPU speed quicker which is good, but it also stays about 500MHz lower maxed out than with it switched off (3.6GHz as opposed to 4.1GHz). That basically negates me getting a NUC this fast. I would very much like to have the Performance profile switch EPP off (if I so choose) so I can make maximum use of the processing power, while selecting the Power Conserve profile switches EPP back on again.
I can do that with the Disable Turbo box just fine so I don't see why EPP is not switchable the same way.

Also, I can't seem to adjust the EPP setting, it's locked at 128...

 

You should be able to click on the number on the main screen and change it, then save it. I use 64, but you can try 0, for maximum.
Good luck,

Joe

 

Hey Joe, I tried it and you’re right. Funny, because I thought it would have to be a box you can click on while it now looks like just some presented unclickable number.

Still, the setting not being tied to a profile issue remains.


Sent from my iPhone using Tapatalk

 

Speed Shift is enabled within the CPU. Once Speed Shift is enabled, it cannot be disabled until you reboot.

The 8559U can run single core tasks up to 4.5 GHz but when heavily loaded, the U CPUs are often times limited by power consumption. You might not be able to maintain max speed when running demanding benchmark tests like Prime95 or any test that uses a lot of AVX instructions. At least your U has a TDP rating of 28 Watts so it should be a lot more flexible compared to the majority of low power U CPUs that are only rated to 15 Watts. Open up the Limit Reasons window in ThrottleStop and keep your eyes open for any box lighting up in red during testing. That will show why your CPU is not running at its full rated speed. A TS Bench - 1 Thread test should show a nice high multi without any throttling.

There is a Speed Shift Max setting that you can adjust in the ThrottleStop TPL window. Set that to 45 and set your Turbo Ratio Limits to 45 for the 1 Core Active setting. A Speed Shift EPP setting of 0 should also help your CPU maintain maximum performance. As far as I know, Windows 10 has taken over control of the EPP setting on some CPUs. That means that ThrottleStop and Windows might fight over control of the EPP value. Use a Windows High performance or Ultimate performance power plan.

Post some ThrottleStop pics and / or some TS log files. It is the only way I can learn about this new technology and make suggestions.

The EPP setting can be adjusted and saved for each profile.

 

Any idea @unclewebb

 

Have you tried disabling HWP in Intel BIOS and set Speedshift in TS to get the CPU back in your control.

 

EDIT4: Update - I think I figured it out (??)... I had copied TS with config from my other computer, and I had max cache ratio at 47. I set it to 43, and the power limit throttling stopped. Either that, or for some reason using the Alienware OC tool and then exiting that somehow changed something. I can now get up to 4.7. Still wondering though - if the limit lights flash yellow/red in the popup Limits window, but Throttle radio button doesn't trigger in the main window, what does that mean? I don't see core speeds drop heavily until the Throttle radio button comes on. So are those popup limit lights just transient limit hits but nothing to worry about, or is that incorrect?

EDIT 4b: well it's not the Cache ratio max, I tried setting that higher again and I am still getting the high performance, 4.7 solid, can set 4.8 but it flutters in the 4.7x range. If anyone has any ideas on what was causing this before, it would be great to hear them. I don't really see how opening then closing the Alienware OC tool would do it. TS seems to always override its settings on my other computer.

Hey guys - so no one got back to me about my last question, I know it was long-winded. Anyway, I just re-pasted, re-padded, and did my extra heat sink/pads on my replacement unit, figured it was the only way to see if the CPU is a better bin than the original unit. I feel like there is some potential here (like I was right - even if it maxes out at same underclock, the adaptive voltage might still let the chip run faster and use less power).

Here's the thing I cannot figure out; I have BD PROCHOT disabled, I have EPP set to 0 and enabled, computer is plugged in (this is an i9-8950HK). I am set to 4.3 Ghz, max temp on any core is 77 degrees, max package power is 83.7W. (the final power limit kicked in on my old one at 108.x watts since 110 is the rated max)

However, even though no thermal is being hit, I still get thermal throttling halfway through Cinebench, all cores slow down to 3.6ish, and I get the EDP OTHER and PL 1 lights going on in the Limits popup.

Anyone have any ideas? I did update the BIOS already, because I know the last Alienware BIOS had issues (this is an AW 17 R5). So the BIOS is the same as my last one. I might not have made all other driver updates but I don't think they would affect this?

I have restarted the computer, so the changes should be set in the CPU for sure, right? As long as ThrottleStop was run the time before? On my other computer, even if it wasn't running, it seems to take effect when I run it, no reboot necessary.

Any ideas? I feel like I am on the brink of some massive power on this second unit.

EDIT: Also, I checked in HWInfo as well, only limits triggered are Power Limit Exceeded on each core.

EDIT2: One other weird thing (well, to me, since I never found the cause of this in my limited searching when first setting up TS) - when just using TS, I always see PL1, PL2, and EDP Other in yellow; then when throttling occurs, PL1 and EDP turn red. Yet when HWInfo opens, they gradually link on and off and then disappear. I think I searched for this one but didn't see it, what does the yellow and red limit button colors mean? Red seems like active throttling, but what is yellow? I thought it just meant what it was capable of monitoring, but now I've seen them turn off completely.

EDIT3: I didn't think PL2 was going off before, but it seems to ALSO be going off. This is very confusing, I know on my last unit, before re-pasting, the heat was what triggered the power limit throttling. Then after re-paste, it was the max 110 watt that triggered the other power limit. Now, it seems both are being triggered, even though I am not going above 92 watts or 80 degrees.

 

Using a separate post for this other simple question I've always had - in the TS docs, it says to fully reset, one has to delete/reset the .ini file, then restart the computer, in order to reset the registers on the CPU. However, I seem to not get the settings working unless TS is running, so it's as simple as TS is running, then it's using its settings. Is this only if one were to set the option "do not reset voltage... on exit" that the deleting ini file, running TS, exiting, and restarting matters? Otherwise, it seems to be that the settings only apply when TS is open and it's turned "on".

Or is that only for desktop CPUs and not notebooks that the delete ini, start/exit TS, restart computer sequence is needed? (or maybe it used to work like that on older CPUs but not any longer?)

 

As long ThrottleStop have loaded your inset settings after boot... Exit ThrottleStop, delete ini file, then restart computer.

 

I know that is what the guide says; but as soon as I stop or exit TS, the clock settings go back to default anyway. So to me, I don't get the need to delete the ini file and reboot; the overclock goes away upon exiting TS. Isn't that option in TS settings "override reset voltage setting upon exit" the one that makes it necessary or not to reset the computer to fully reset the CPU? It makes sense to me if, only upon choosing that setting, that one would need to do the delete ini, restart TS, exit, reboot computer sequence.

 

I don't know all the many ways Dell's engineers crippling their firmware, but last used TS settings should still be there after you you exit TS. Only after restart and without TS load up you should see default settings.

 

@unclewebb - I know you're busy, but is there an additional guide somewhere to learn about exactly what registers get set and unset even when TS exits? As mentioned above, it's very confusing to me, because as soon as I click Stop or exit ThrottleStop, the overclocking is no longer in effect. It's the same with Intel XTU. The last time I seriously overclocked a computer in 2011, while I used BIOS for final settings, from what I remember, the software to overclock also only was in effect while it was open. Software closed, no more overclock.

So is it that all settings should be kept in CPU, including max ratio per core, undervolt, etc. (and it is indeed limitation of Dell laptops that prevents that from working as expected), or is it just stuff like the SpeedStep, C1E settings, stuff like that?

From another perspective - if the settings persist (or are supposed to) when ThrottleStop exits, why in the world is there a guide to have it auto-start with Windows? I thought that was so that it would automatically open up and apply one's overclock and other settings, no? But if it's all supposed to be stored in the CPU somehow even through reboot (this part I really don't get btw - there is permanent storage on the CPU to store overclock settings? that should all be BIOS? And Bios settings can't be touched by TS or any other client software, can they?), then why is there a demand to have TS open automatically upon Windows login?

As you can tell I am really confused on this.

Same confusion on how one OC tool affects another - such as with using Dell overclock built-in tool, and ThrottleStop - it seems apparent that TS overwrites any settings that the other tool had set. Yet the guide mentions closing the other tool, otherwise TS will treat that setting as "default" and build from there. But I don't see that happening. Otherwise, shouldn't my -100 mV max undervolt in the Dell tool be added to the -125 in TS, and it would be -225, which is guaranteed to crash? Yet what happens is, running the Dell tool then TS, or just TS, yield the same overclock results. (i.e., only TS's settings take effect, the CPU is clearly running at -125 mV undervolt in total, not -225 mV)

Thanks. If there is some other thread or guide, I'd love to read, but most threads I come upon seem to be specific to individual's builds. If there's a more in-depth guide to the Limits pop, and what the yellow vs red means (vs not showing at all), that would be wonderful as well.

 

@unclewebb

Ok so my friend got an X1 Extreme with 8850H

NBC review: https://www.notebookcheck.net/Lenov...GTX-1050-Ti-Max-Q-Laptop-Review.335608.0.html

With the 8850H and TS overclock enabled (to change the CPU multiplier and get higher speed) it gets severe current limit throttling regardless what setting is changed/tweaked. CPU isn't nowhere near overheating and is actually staying below 45w.

Apparently Thermal Velocity Boost is causing the CPU to current limit throttle, only able to get around 870~ score on CB15 CPU. This triggers at 53c (50 is Intel specified).


If overclock setting is disabled on TS, the CPU is able to stay at 3.1GHz 55w under 80c and get the expected 1200 score from CB15.




So when overclock setting is enabled (to enable change of the multiplier) it's accessing something and at the same time it's activating thermal velocity boost.

Is it possible to get TS to be able to change multipliers without triggering TVB? Or make it able to have control over if TVB is enabled or disabled?



@Papusan @bloodhawk @Mr. Fox

 

You are right. Almost winter here. Not much time for TS or the forums lately.

If you are only using ThrottleStop and do not have XTU or some other Dell software running in the background then your overclock and voltage settings should still be in effect after exiting ThrottleStop.

ThrottleStop writes multipliers and voltages to the CPU but these changes are not permanent. As soon as you reboot, the bios is supposed to reset all of these changes back to default settings. The moment you exit ThrottleStop, Windows or any other software running on your computer will be able to change these settings. When ThrottleStop is left running, it actively monitors the settings that you have selected within ThrottleStop and tries to maintain those settings. If XTU and ThrottleStop are both running at the same time, keep an eye on the TS data to determine what program is in charge of your CPU. Some of the XTU data is not updated on a regular basis so changes made to your CPU by TS will not be detected or reported by XTU. It is not a problem running both programs at the same time. It is just not recommended because most users put too much trust in XTU and what XTU is showing them when really they should not.

Undervolt settings are not cumulative. If you set -100 mV in XTU and -150 mV in TS; the CPU will mostly use the -150 mV value because TS is a bully. TS tries real hard to be in control of your CPU. That is why this thread is 900+ pages long. Long time TS users know they can generally trust TS and what it is telling them.

When you run TS for the first time, it specifically tells you that TS Limit Reasons is not compatible with HWiNFO. The data in the CPU that keeps track of throttling is continuously reset by HWiNFO. Those yellow boxes in TS show previous throttling episodes. If HWiNFO is running in the background, it will constantly clear this data from the CPU.

That is not thermal throttling. EDP OTHER is usually the current limit and PL1 is the turbo power limit. A CPU can throttle for a wide variety of reasons. Thermal is just one reason. If Limit Reasons does not show THERMAL lighting up then it is not thermal throttling.

Are you using the FIVR - Disable and Lock Turbo Power Limits feature? I highly recommend it.

I know the recent versions of Windows 10 on the newer CPUs will change the Speed Shift EPP value based on what power profile you are using. I do not know how to block this from happening.

No 8th Gen hardware to play with for testing purposes. I have never heard about Thermal Velocity Boost and I have no idea how to control it. Maybe someday.

TS seems to be doing OK on the 8th Gen desktop CPUs without any TVB going on.

http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-1662#post-10813507

TS is also working OK on this 9600K which was news to me.

http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-1660#post-10813297

 

The answer is easier than you think...

Spoiler: Open for the answer...

DON'T use a crappier recent version of Windoze OS X that causes that.

Use one that does not and block the sucky cancer updates.

 

Tell your friend that he has my deepest condolences. Maybe he can return it for a refund and get something decent.

 

TVB is only on mobile iirc.


If you need data to investigate I can provide them.

 

@unclewebb sorry I meant 3.1GHz is the current throttling thing. If overclock is disabled and TVB is inactive it'll go higher than 3.1

 

@unclewebb Thanks so much! That really clears up a lot for me. I had a feeling HWInfo was playing with the flags and making those limits go crazy. Yellow was throwing me for a loop because sometimes it starts off yellow, but it makes sense now that it means in the past. I will definitely try the Disable/Lock feature next. I was going to ask about that, eventually, but it must be time now! Thanks again.

 

I read this topic
https://forums.anandtech.com/threads/what-controls-turbo-core-in-xeons.2496647/page-98

And people on Xeons can mod their bioses to change multipliers to maximum on 4 cores.

My problem is that i have QDE4 engineering sample and it is 4770K.

But my motherboard Gigabyte B85 D3H v1.0/1.1 is blocking to overclock it and i can only force max turbo. I can't even change voltage.

Default clock is 2800Mhz and i can force it to 3200Mhz on all cores.

Can anyone help? Maybe we can change microcodes or something.

I tried flashing older bioses but with no profit.

This is my CPU-Z report:

https://www.sendspace.com/file/iphms2

 

Thanks for your answer. Then do you know how I can check what's my current EPP? So I could see what has the priority: your soft or Windows. Or the last one which edited the settings. Thanks.

 

The table in the top right corner of the FIVR window shows the current Speed Shift EPP value that the CPU is using. The data in that table is being updated approximately once per second.

Both Windows 10 and TS are writing data to the exact same CPU register to control EPP. I do not know the frequency that Windows is doing this. Recent versions of TS might be winning the fight over this register.

The B85 chipset might not allow any CPU overclocking. In that era, the Z87 chipset was the one that Intel designed for overclocking.

Thanks for introducing me to Thermal Velocity Boost. Unfortunately there is probably not much I am going to be able to do until I have some hardware to play with and some documentation from Intel. That might be years from now. Are you using the FIVR - Disable and Lock Turbo Power Limits feature?

 

TVB is only Supported on Intel® Core™ i9-8950HK and Intel® Xeon® E-2186M. <Intel> implemented this uselsess feature (I call it scam) to be able to charge $200 more than usual from the OEM's/Notebook manufacturers. And as usual the notebook manufacturers continue the scam and add $600 on top vs. H mobile cpu's if you jump on this...

♫ The Core i9 clock cycles go up. Who cares where they come down?
That's not my department say, er... Intel, Apple and Dell ♫

Owners of laptops fitted with Intel's Core i9 high-performance processor, including computers made by Apple and Dell, are finding that the machines slow down compared to the pace of older models.

 

FTFY, @unclewebb

 

No, I'll check it after this.

Do you need any experimentation data from the X1E that will help you try and investigate this?

From the testing and experience with X1E it does seem that TVB is somewhat active or partially active on the 8850H.

 

8850H is partially unlocked filthy. It is possible some Notebook manufacturers/OEM's has taken some shortcuts. Official info from Intel https://newsroom.intel.com/wp-conte...8th-gen-intel-core-h-series-product-brief.pdf

 

My B85 can overclock 4670K but QDE4 not.

 

TVB is pointless - I don't get the point about trying to optimize it in TS? It only kicks in, what, below 50 deg C? So basically if the computer is doing absolutely nothing, and you need a 3 second spurt, it might help, right. (maybe 1.5 seconds actually - as long as it takes for the temp sensor to register it) I really don't get it in a laptop. Desktop, sure, give some extra kick if the chassis cooling is working really well.

 

Thank you so much for your help.

Here is my help in return:

I've played a bit with both my ThrottleStop Profiles and with default Windows Profiles.

My TS profiles are :
1: 0
2: 96
3: 192

According to the FIVR window, they are ALWAYS applied when my windows profile is set to "best performance".

According to the FIVR window again, they are NEVER applied when I choose any other windows profile.

Best battery life / better battery life: 178 EPP
Better performance : 127 EPP
Best performance: TS profile currently set. (FYI I haven't checked what EPP Windows chooses in case TS is not running).


So conclusion: if you want TS to work, just always pick the "best performance" Windows profile.


It would be interesting to check on other PCs to make sure my statement is right but I don't see why it would be different. So feel free to add my conclusions to your OP so people don't get disappointed by your great soft'.

 

I just wanted to make a point on the Windows power profiles - in Windows 10 anyway, there is just the balanced plan by default. The names "performance", "balanced", etc., are just that, names. Some OEMs might add more by default, I don't know. You can edit any plan to make the CPU behave how you want, you just have to go to Advanced Plan settings, drill down into the CPU area, and set the min and max, set the power savings mode for the CPU, etc.

But yes, any overclocking (for performance reasons) should always be done plugged in and on the best plan settings, whether there is an existing default high performance plan, or copying and making a new plan (and calling it high performance or whatever one wants).

 

Hi! Thanks for ThrottleStop.
It's very useful and I enjoy reading your various technical responses here.

I use TS 8.60 with an XPS15 9550 (i7-6700HQ + 960m).
Undervolted by -150 on CPU, cache and GPU.
Speed Shift enabled (1 ... 35).
Speed Shift - EPP (ON) 128
SpeedStep (ON)
C1E (ON)

For a long time I had obvious throttling, visible under the Limits window.
It had EDP OTHER tripped all the time, frequently would trip PL1 and PL2 etc.

I repasted it and applied thermal pads per common guides to great effect.
After doing that the Limits window is clean all the time, even EDP OTHER stays gone.

Perfect except ....
Randomly while gaming, the clock multiplier will drop all the way to 9x and the system nearly halts being very unresponsive to basic tasks. But the Limits window is still empty, the Throttle and PROCHOT 97c aren't tripped either. No apparently reason for why the multiplier dropped to 9x.

While it's doing this if I set Speed Shift - EPP to 0, the multiplier goes back up to 32 and the system is responsive again. The game still feels wrong though, like the system is still throttling somehow. If I set that back to 128 the clock will go back to 9x immediately.

The particular game I am playing drops to 15 FPS when the window is inactive, so I can easily test reducing the CPU/GPU load just switching away for a while. After 5-10 minutes switched away like that this 9x multiplier thing goes away.

Additional things I have tried:
Get rid of Internal Dynamic Thermal Framework.
Enabled Ultra Performance thermal profile in Dell Command Power Manager ("processor and cooling fan speed is increased for more performance")
Set windows power plan CPU min/max to 100 and turn off any power saving looking options.

Any ideas what sort of limit might be tripped that ThrottleStop is not showing?

 

It sounds like your GPU is the device throttling, and the CPU slows down simply because the game has slowed down (due to GPU), hence no work for the CPU to do. That fits with raising EPP to 0, because that will force max clock all the time. This is assuming you've also logged CPU temp and it's staying below thermal thresholds (not sure what it is on your chip, but let's say less than 90 degrees C?). I know thermal alarm hasn't gone off, so it sounds ok, but just to be sure. Unless the CPU load showed 100% when it stuttered, was it?

I actually had the same issue when changing my panel to a 4K from a QHD screen, game would stutter every time there was action. (AC Origins, all ultra settings) Lowering settings didn't help much. I wasn't as knowledgeable about all the limits, but CPU temp wasn't above 90 max. Anyway, when I did a re-paste with liquid metal, suddenly ultra settings at 4K worked fine, 40-55 FPS. I can't explain it to this day. GPU wasn't going over 75; CPU wasn't hitting the threshold for thermal throttling. I had already re-padded the VRMs on the motherboard. But somehow changing to liquid metal did help.

Sometimes I have to wonder if the temp sensor isn't the full story; maybe the average heat throughout the chips matters too. So the max temp looks fine, but if the average temp of the silicon is also that high temp, maybe that is no good. I would assume the temp sensor in the chip is engineered to capture it in a way that is useful, of course. This is just some far out speculation on my part.

Are you running on a cooling pad btw? Have you tried forcing more air at the intakes just to see if that makes a difference, or running with cover off, etc.?

 

I thought of this, but it didn't seem to fit that Windows itself would be nearly halted while at 9x multiplier.
Also, if the GPU was limiting so much that it causes the game AND windows to nearly halt, so the cpu multipler scaled to 9x with nothing to do, then forcing the CPU multiplier up via EPP=0 or a direct multiplier setting would not change the GPU limit that set off the chain of events. i.e., if the GPU was the source of the problem it would not go away regardless of the CPU multiplier.

The max CPU temps reported by TS seem ok, max ranges from 75c to 86c. (That's 25 and 14 as DTS, still a ways off from 0)

In this case the game runs buttery smooth for hours, and only has this issue sometimes.
Seems related to ambient temperature if I had to guess.

Yeah, it certainly seems thermal related. What's odd is that there is no visible throttle indicator, and the CPU lets me force the multiplier up so the thermal situation is clearly not that bad.

I use it on a hard surface that allows good airflow to the bottom intakes.
Never actually tried one of those cooling pads.

 

I see, I missed the part about Windows being unresponsive also. Video glitching is one thing like if it freezes, but if it's truly running slow, then yeah. Plus I didn't realize you were saying the EPP change also improved the responsiveness.

I think the cooling pad is a necessity if one's going for highest performance and using GPU and CPU at once. Even just lifting a notebook more in the air seems to help, I had a Macbook Pro that was like that. Not even a fan stand, just a riser, and the internal fans didn't work as hard. I am pretty sure I've noticed a few degrees difference on this Alienware even just turning the pad's fan on and off while the laptop is under load. Might be worth a shot (or at least lift, and maybe run with cover off - if you have any kind of fan, just blow it at the bottom/intakes). Keep track of ambient temps too - sometimes I am testing and temp is markedly better or worse, and then I realize from my wall thermometer, oh that's why, it's only 71 F (or it's 80F).

Hopefully others will have an opinion too!

 

-150mV on the iGPU is way too low for the XPS 9550 and it might be the root of your problems, a heavy iGPU undervolt usually won't cause a bluescreen like a CPU undervolt would, but it will freeze your system under certain loads (as the Desktop Window Manager relies on the iGPU not on the dGPU). I personally use -100mV on the iGPU and everything is fine.

You could also dig into your Windows power profile and disable every power saving feature that might be enabled.

 

I can confirm too. I used to have high undervolt for iGPU and experienced random freezes while playing games. I checked event viewer and found many warnings related to igpu driver reloading constantly. I turned off iGPU UV completely and never faced any such issues later.

 

Hello guys! Fist of all big thanks to unclewebb for this amazing software.

I have few questions (I use win10 + Intel 7700hq + gtx 1060):

1. When I turn on Speed Shift - EPP, graph of cpu frequency (In ETU) has become very jerky than before. It's ok for processor?

2. When I use Speed Shift - EPP do I need to turn off SpeedStep or something else?

3. I need to make three profiles for "Max Performance",


"Lowest Performance"


and "auto"


how I can do it by best way? I do it right on screenshots? Now I have it like this, but I don't sure.
For low and max performance I think it's ok, but how I can make Speed Shift - EPP in auto mode or something like this?
Or how I can turn off TS? Even if I click "turn off" button, Speed Shift - EPP work anyway.


Thanks for answers in advance.

 

Most mobile Core i processors are able to change their multiplier and CPU speed hundreds of times per second. This is normal. Most monitoring software is only sampling your CPU speed once every second so it would not be unusual for a graph to look like a saw tooth or very jerky. No worries.

When Speed Shift is enabled, the SpeedStep setting on or off does not matter. The bios on many computers likes to enable SpeedStep by default so if SpeedStep is checked, just leave it checked. Ignore it.

When using Speed Shift, the old methods to control your CPU are ignored. That means if you see SST in green on the main screen, Speed Shift is enabled so the Set Multiplier value is ignored. Do not check Set Multiplier when you are using Speed Shift. It does not do anything.

The Power Saver option in ThrottleStop is SpeedStep related. Do not check this option if you are using Speed Shift. This option is for 10 year old CPUs. Not relevant for the newer stuff.

I rarely see any need to check BD PROCHOT, especially when plugged in. Some people like to check BD PROCHOT when running on battery power. That is fine if it does not cause any throttling problems on your laptop.

For maximum performance when plugged in, use a Speed Shift EPP setting of 0.

When running on battery power, a Speed Shift EPP setting between 80 and 128 works well. If you want to keep things simple, a single profile with EPP set to 80 is OK whether plugged in or on battery power.

If you want a slow and sluggish computer you can use an EPP setting of 255. My question is why? This will definitely make your computer run slow but will it save any power? That is debatable. When a CPU has something to work on, it does this most efficiently when running at a fast speed. When you force a CPU to run at a slow speed, it can end up consuming more power so you are not accomplishing anything. I love a fast computer. I hate slow. That is why I wrote ThrottleStop.

The Turn On / Turn Off button only applies to the Set Multiplier and Clock Modulation features. You are not using either of those options so Turn On / Turn Off changes the color of the icon in the system tray but does not do much else besides that.

You are welcome. I am always happy to hear from new users that have just discovered ThrottleStop. It is hard to imagine what a sorry state the laptop industry would be without the help of ThrottleStop. All manufacturers have built a butt load of miserable performing laptops in the last decade. TS to the rescue!

If you are having problems like this the best thing you can do to help me is post a ThrottleStop log file somewhere so I can have a look. Also post a screenshot or two of TS so I can see how you have the program setup. Before you check the Log File option, turn on things like Nvidia GPU monitoring and Add Limit Reasons to Log File in the Options window. Your problem almost sounds like Clock Modulation throttling. Is that checked and set to 100%? Dell used to like using this "feature" to kill performance.

 

For those who use DPTF. I have found new version on windows update catalog, version: 8.5.10101.6917
http://download.windowsupdate.com/c..._a17a539098f4913b9c96a09eb571609a3bddd7b1.cab

 

I have been running with Clock Modulation (OFF) 100%.
I'll turn it on and see if I can reproduce with everything else the same.

I've never seen the Mod column on the main TS window below 100% while experiencing the 9x multiplier problem, if that's meaningful.

This is the configuration I had been using prior to suggestions from this thread.

 

is the iccmax value supposed to be the same on cpu core and cache?

 

Hi, I have a Zephyrus M with the 8750h cpu, and I'm using ThrottleStop to lower the insane stock temperatures. As almost every gm501 owner already did, I managed to undervolt -125mV / -48.8mV on core and cache respectively and get some temp drops, but I still have some questions, I hope you could answer me.

1) The gm501gs has a very tight undervolt margin unluckily, but I noticed that if I push cache over the safe limit of -48.8mV (-65,5mV, for example) I still don't get any crashes during both standard use and games. If I run TS bench with that setting, though, I get a lot of errors neverthless of the type of test I picked. What should it mean? Is that undervolt setting "safe" or not?
Also note that, sometimes, I get 1-2 errors max if I do the heavier test, using the -48.8mV cache setting.

2) In TPL section, what should the best settings be for turbo long and short power limits? These are my main TS and TPL window (the TPL ones are the default values since I removed the .ini file):



And what does that ticked "clamp" option do exactly?

Thanks

 

Try to stress test the laptop with software that stress both CPU and GPU like Rog realbench (free) or 3DMark. I was confused when I can undervolt my 6700HQ by 0.151mv testing with prime95/linpack xtreme/OCCT but crash immediately in games. After a lot searching, someone said GPU load decreases the voltage of CPU when the power is limited. Also try to make sure your CPU isn't throttled during stress test otherwise it's useless if your CPU stays at base frequency.

 

Dears,

There is a very problematic interaction between ThrottleStop and my laptop battery, running Windows 10:

1. If throttlestop is ON, and I unplug my AC, my laptop freezes;
2. If I am on battery and plug the AC, my laptop freezes when I open throttlestop;
3. If my battery is not fully charged, and I open throttlestop my laptop freezes again

I am using balanced power plan, also tried with performance plan, also set minimum processor state to 100%. Nothing works. Once, I had "whea uncorrectable error" blue screen on Windows 10.
Any suggestion? @unclewebb ?

I am using a ThinkPad E480 with AMD hybrid graphic card (i7 8550u, RX 550). Here is my ThrottleStop.ini: https://pastebin.com/jehnHYY7

 

That is a pretty good sign that your laptop is not stable. You are using an undervolt of -125 mV. Why? What sort of stability testing have you done to confirm that your undervolt is stable at full load and part load? An undervolt setting that is stable when loaded on AC power might be too aggressive when running idle or at a low multiplier while running on battery power. A low power U CPU is already designed to significantly lower the voltage when idle. Throw a big undervolt on top of this and I am not surprised that your laptop is freezing.

ThrottleStop is not the problem. It is your settings. Exit ThrottleStop, delete the ThrottleStop.INI config file and start again with zero undervolt. Do not drag the voltage sliders randomly to one side without doing lots of testing along the way.

I don't think it matters. Why not just set both of them to 255.75 and call it a day.

Why not upgrade to TS 8.70.6? Download link is in my signature. It has another option or two that might help you.

I think on the 8750H, the CPU core and cache undervolt amounts need to be set equally. Any errors at all in the TS Bench test are a bad thing. Why would you trust a computer if sometimes 2+2=4 and other times, 2+2=5? That is what a TS Bench error is. This is your CPU core or cache literally screaming for some more voltage so give them some.