The ThrottleStop Guide

The Windows Event Viewer has lots of info in there.



That tells me that I should probably clean out my old logs more often than I do.

The default voltage that Intel uses is always too high. It has to be to cover their butts. Most processors need slightly more voltage as they age so Intel tends to set the default voltage a little on the high side just to be safe. Not sure why but on Skylake CPUs, they have taken this "a little extra voltage" way too far. I think my Skylake desktop Core i3 can run stable with 1.08 volts at 3.7 GHz. If I let the CPU manage itself, it ends up with ~1.25 V. That amount of extra voltage is totally unnecessary. All that does is it wastes power and creates extra heat. If you are using ThrottleStop, might as well fully use it and trim off some of that extra voltage. On locked CPUs, reducing power consumption allows the CPU to operate at full speed longer before throttling. It is a win, win, win situation.

 

Ok.. so I've got it launching fine now. I goofed and had "Task Bar" unchecked.
I am launching via Task Scheduler, exe located c:/TS/Throttlestop.exe

Problem I have now is I can't get "CPU Temp" to check. It simply doesn't check, but "CPU Mhz" does?
I've continually done Icon cache refreshes.
Interestingly Windows 10 doesn't have Throttlestop listed as an Icon I can show in the Notification Settings you screen grabbed above. Likely this is part of the problem, but not sure how I can get it to show.

 

ThrottleStop 8.40
https://www.sendspace.com/file/p6wkzr

New Features
- add Speed Shift Energy Performance Preference (EPP) requests for each profile.
- add realtime EPP reporting to the FIVR monitoring window.
- add Speed Shift (SST) reporting to the main screen.
- added Stop Monitoring option to reduce CPU usage when not needed.
- add C8, C9, C10 C State reporting.
- fix C10 request reporting.
- fix ThrottleStop randomly popping up.

 

now that you mentioned this i should revisit my xps m2010 and see if thats causing my cpu to throttle but it probably different scenario than the one you stated here. remember i msg you a long time ago overclocking T7600G in that laptop would cause clock mod and chipset clock mod to decrease to just 12.5% by itself and even if i have those boxes checked set it to 100% it doesnt help. do you know what might be the cause of that?

i just mostly disable c-state and set two different profiles. in use c-state off, not in use cstate on, power saver on, low frequency to reduce power consumption.

also any chance for 5th profile coming? and turbo boost 3.0 haha

 

Hi all

I'm getting 'Residency' reported as a limit reason - what does this mean?

 

For my resource-intensive real-time music software, disabling C-States made a big performance difference but...on my Skylake processor, disabling C-States prevented the processor from reaching max turbo speeds. I found a simple work-around.

Disabling just the "problematic" C1E states solved most of the music software performance issues (latency and dropouts from CPU) AND allowed the processor to reach max turbo speeds. Best of both worlds.

For my Dell 9550 max performance requires me to:

1. Enable C-States in BIOS
2. Then in ThrottleStop: disable C1E (check box), enable SpeedShift (check box and type 0 in EPP box)
3. Turn on ThrottleStop (button on main screen will say "Turn Off")

___
FYI -

A few new computers might have SpeedShift and EPP support in BIOS or Windows but most do not. If you can set SpeedShift and EPP in BIOS give it a try; otherwise you can try with ThrottleStop

I enabled SpeedStep in BIOS and disabled SpeedStep in ThrottleStop (uncheck box) for some expirments, although Uncle Webb notes below those steps are not necessary.

Also M$ reporting of these items is not reliable (via powercfg or control panel>power options). ThrottleStop will show a green "SST" code on the main screen when SpeedShift is enabled. I am also seeing the correct current EPP level reported on the FIVR screen.

 

There is Intel's official definition. Maybe you could ask Intel to translate that to English. I have read a lot of Intel manuals but that definition wins the prize of being the most obscure definition ever. I don't have a clue to what that means. Does this just happen when your CPU is idle? Not everything that Limit Reasons reports is a bad thing. Some stuff might just be an indication that a power saving feature has kicked in. Is this showing in red or do you just see it in yellow?

Any chance I can have a holiday first? There is always something that needs improving. If the new Kaby Lake CPUs support Turbo Boost 3.0, that might give me an excuse to buy one. Maybe next year. Send me your T7600G and I will have a look at it. Not sure what is causing that issue or if it can be fixed.

@pressing - When using Speed Shift, I don't think it matters whether SpeedStep is enabled or not. Speed Shift should take priority over SpeedStep. Speed Shift takes over control of the CPU speed so writing multiplier values to the old SpeedStep register gets ignored. If you need to control your CPU speed, Speed Shift is probably a much better way to do this.

 

I corrected the text.

FYI - There were early tests where enabling SpeedStep in BIOS helped maximize performance so users can keep that in mind when experminting. I can't remember details unfortunately...

 

I see it when running various benchmarks, and I've seen it in red during the built in 1024 benchmark in ts. So I wouldn't expect it to be a power saving feature, maybe it's reporting power limiting?

 

The pp0 limitation on my device is so annoying--even though I don't expect some fancy performance on a tablet, the 12A current limit makes it a bit awkward anyway. I indeed saw MSR 0x601 has 12A limit with MSB locked. I am wondering is pp0 something controlled in the BIOS? I have seen TDC current limit and VR current limit in the BIOS hidden option (I cannot read their values though), and am wondering if these have anything to do with the pp0 current limit.

 

Hello!

Firstly thanks for the great app!

I have one little question.

My laptop uses i7-2620M CPU. According to specifications turbo speed of the CPU should be 3.40 GHz, with TDP 35w:



But it seems that I cannot pass over 3189.69 and 26.6W. What am I doing wrong? Is there any workaround?









P.S. I have advanced tab unblocked in my laptop's BIOS.

 

Interresting info as usual Do you know why CPUs require higher voltage as they age? Will running them at lower voltage reduce that? I get -185 on my skylake CPU, if it requires slightly more after some years it dosent bother me, but if you know if voltage itself contributes it would be welcome info

 

There's a billion+ transistors on each chip ... in theory. Even when brand new not all will be functional due to flaws in the manufacturing process (better chips from the heart of the wafer become an i7 and worse ones a celeron, eg). With use more and more of the transistors will die, part of them left 'On' all the time, which increases overall resistance yet decreases the number of operations the chip can handle. The result is lower efficiency; not all power put in will actually be used since there'll be 'vampiric' transistors. Much like overhead in overly large organizations, really .

There's also a large number of small capacitors inside (mostly for the cpu cache), which are equivalent to batteries. As with all batteries these, too, will age and their internal resistance will increase, requiring more energy to charge fully or the same amount of energy for fewer working cells, lengthening calculation time. The same problem happens with ssds (which are basically enormous battery banks), especially with triple-level cells; the difference between 00, 01 and 10 and so on will become ever harder to determine. This was the root problem with the Samsungs; the firmware didn't compensate automatically for this voltage drift.

Don't know which effect is more pronounced with cpus, though. Either way, lower temperatures will help, of course. Then again, is it worth the effort? You'd have to do proper research to measure the effect. After all, if we're talking about a few percent after years and years of 100% high-temperature usage then there's no reason to bother with conservation measures. Less heat, lower fan noise and more battery time are much better motivators (performance will not suffer). Or naturally, if the system permits, higher frequencies; might as well use the otherwise redundant safety factor Intel sets for each chip (ymmv).

 

I saw that there is new version 8.40 download at the first post,

ThrottleStop 8.40
https://www.sendspace.com/file/p6wkzr

However, the downloaded file is not valid, the file size should be 6xxKB but the actual file size is only 530KB. It cannot be opened by WinZip and it shows the file is invalid. I tried several browsers to download but it is no luck. Any idea in it? Does the file get damaged? Looking forward to trying the new version. Thanks a lot.

 

I just clicked on the link you posted and TS 8.40 downloaded fine, and all the checksums I ran matched the original zip file I uploaded. You can try downloading it from my Google Drive.

https://drive.google.com/open?id=0B0dpSo9k93jDcUc0QkxOLV9RUGM

Edit - Are you running an overly eager virus scanner? That might prevent the download from completing. Windows Defender was causing issues last time I released an updated version. It might take a few days before it realizes that TS 8.40 is not that bad. VirusTotal is happy so far.

http://i.imgur.com/lpjpn4p.png

For Intel specs, I like using CPU World.

http://www.cpu-world.com/CPUs/Core_i7/Intel-Core i7 Mobile I7-2620M FF8062700838809.html

That says the 34 multiplier is only available when a single core of your CPU is active. When both cores are active, the maximum multiplier drops down to 32. To access the full 34 multiplier, when a single core is active, the second core has to be inactive and it must go into one of the deeper C States like C3 or C6. Some people love turning off the C States but on most of the non-K CPUs, you will lose access to the maximum multiplier if you disable all of your C States. Thanks for posting those pics. I saw lots of 0.0 in your C State window so I am assuming that you disabled these in the bios or in Windows.

Your Core i7 is a Dual Core CPU with hyper-threading so it is not overly power hungry. The TDP rating is a maximum rating. If your CPU is able to run at its maximum rated speed without reaching the TDP limit, that is a good thing. Did you try running Prime95 Small FFTs? That is a good test to show maximum power consumption. LinX with the latest Intel binaries is another good test and might show slightly higher power consumption compared to Prime95.

stasio at TweakTown does a great job maintaining his list of overclocking and monitoring tools.

http://forums.tweaktown.com/printthread.php?t=30530&pp=10&page=1

Usually the bios sets the current limit in MSR 0x601 and then decides whether to lock it or leave it unlocked. For most laptops and tablets, this setting is going to be hidden. You might see different names for the current limit but usually, they are all referring to the same MSR register.

I guess that was wishful thinking. When this lights up in red, post a picture of ThrottleStop and Limit Reasons while you are testing. Show me the TPL window so I can have a look at your power limits, etc.

I have seen some bios versions on a few different Asus boards where the SpeedStep option in the bios didn't work. Sure there was an option in the bios to enable or disable SpeedStep but after booting up into Windows and checking, SpeedStep was always enabled. I think they did this deliberately to prevent the CPU from getting locked into a low multiplier state. Not sure how many other manufacturers use phantom options like this. As soon as Speed Shift is enabled, it should be in full control of your CPU speed.

 

@unclewebb

v8.40 is uploaded on Major Geeks = http://www.majorgeeks.com/files/details/throttlestop.html

Thank you sir

 

The download worked fine for me, it may be your anti-virus: http://i.imgur.com/iB0PNPa.png

 

Hey, @unclewebb
Awesome utility. I was able to reduce the temps on my processor by about 10-12C

But, I'm unable to minimize the program to the system tray (tried 8.30 & 8.40), is this a known bug? or am I doing something wrong?

 

hopefully the image works! If not here's the url http://imgur.com/a/PWr5t

It seems to be triggered by heavy duty benchmarking ( I ran the ts 1024 benchmark to trigger it this time, but firestrike and passmark will do it as well), but once it's happened it can be set off by something much more benign, and will even kick in a bit at idle.. If I let it idle for a while it stops happening. When I took that screenshot the multiplier was set to 46 but as you can see it was throttling to 41.

 

Thanks for sharing.

Yes, this is a known bug but there is a solution. You will need to clear your WIndows Icon Cache. Details here.

http://www.digitalcitizen.life/how-clean-notification-area-icon-cache

After you do this, only run TS 8.40. If you go back and run different TS versions, you will need to clear your icon cache again and again and again.

Edit - I just noticed your screenshot. You have Task Bar checked? That makes TS minimize to the task bar. Clear that box to minimize to the system tray / notification area.

@button22 - A link to an image file will end in .jpg or .png. Those links you are posting are not links to image files. That is why they do not show up in forums. Click on an image in imgur and find the Direct Link. Here is an example.

http://i.imgur.com/KuM7x6s.png

Unfortunately, I don't have any experience yet with RESIDENCY related throttling. Perhaps @Mr. Fox or @Papusan have seen this type of throttling in action before. They are the TS - 6700K overclocking experts around here. I am just the programmer and a decent Skylake or Kaby Lake K series CPU is still in my future. I have a couple of minor suggestions you could try. In the Turbo Power Limits window, never use the Clamp option. Leave that box clear. On the main screen, do not check the BD PROCHOT function. Try checking the Set Multiplier option and set that to 46. It probably won't make any difference. Can you also post a pic of your FIVR window?

Edit - Have you ever tried controlling your CPU with the Speed Shift function? Try enabling that and set EPP to 0. Do another TS Bench test and see what the multiplier does.

You're welcome. Happy to hear that you are having fun with ThrottleStop.

 

Fortunately, I have a @Prema BIOS mod and I am able to disable Residency State Regulation. Most Haswell and Sky Lake notebooks are cursed with horribly poor optimization for stock CPU performance with voltage jacked way too high for proper function. Even before we had the ability to disable RSR with Prema's mod Current Limit Throttling was a much bigger issue. If someone is having throttling problems I would look there first before blaming RSR. HWiNFO64 may show RSR as a performance limiting reason, but even then that is usually a side effect symptomatic of Current or Power Limit throttling.

Throttling can often be reduced through fine tuning voltage unless the OEM has baked in their cripple code too deeply to correct the problem with custom tuning. The Limit Reasons tool @unclewebb includes in ThrottleStop is very useful in tuning performance, as you can use it in conjunction with the benchmark tool to find the sweet spot for load voltage. The 1024M test runs long enough to give you time to tinker and watch the effect on clock speeds and throttling behavior in real time.

 

Thanks for your help. Yes, I can get it on your link.

 

Hello,

I have an XPS 9350 laptop and am using throtlle stop to under volt (-65mV) the CPU. This allows it to run full speed all the time without overheating.
Without it, running compilation code under a vm always make it reach 100C triggering throttling.

Now I am trying to understand how to replicate those undervolting under linux.
I understand that the voltage offset setting is not using the MSR register but something else but after reading tones of materials (intel CPU spec, motherboard manual for voltage settings) I can't figure out how the voltage offset is passed by to the hardware from the operating system.
It seems ThrottleStop use the winring0.dll driver to access the ACPI function that expose this capability but I can't find any table or document explaining which function do that.

I also took a look at the phc code used to do voltage regulation but this is for older CPU using ACPI calls to modify the frequency/voltage table.

Can someone point me to a document explaining which function you are using in the winring0 dll to change the voltage offset of each CPU element?

Thanks,
Alexis.

 

@button22 - I found a bios option for a Gigabyte board that has a much better definition of Residency.

If I ever find a way to disable this nastiness, I will add a new option to ThrottleStop. If anyone has a motherboard with this option in the bios and you want to help me do some testing, send me a message. I cannot remember seeing this option in the bios of my low end Skylake desktop board.

@Alexis Boutillier - Unfortunately, adjusting Intel CPU voltages is not a secret that can be shared in a public forum. You need to sign an NDA with Intel. I am not willing to share this information, publicly or privately. It is not worth the risk.

 

That would be great if you could add that for folks that do not have the option. Everyone that has a Sky Lake Clevo with @Prema BIOS has that option in the BIOS. I know that I and @Papusan could help you with that. Probably @Phoenix, @TBoneSan, @Johnksss@iBUYPOWER, @Meaker@Sager and others would not mind helping test that new feature as well. Do you need a dump of some kind to identify the MSR address of the code?

Question: Do you know what IMON Scaling Support does or what it is for? I have not been able to find a legitimate technical explanation of that on the web. Most motherboard user manuals (including the best of the best) are totally worthless when it comes to providing meaningful explanations of BIOS menu items and features.

 

Thanks for the info - I can't quite see why that would be any better than ordinary thermal throttling but I suppose there must be some reasoning behind it. I have seen a post somewhere (from Mr Fox I think), saying that he always disables RSR in his prema bios.

I have in fact undervolted my settings for 4.6 gig a little and the problem seems to have gone away, which chimes with that info. Still no luck with 4.7 though, it seems to need the extra volts to be stable. So I guess the only other thing to do would be to work on the cooling...

 

I checked my Desktop board (Rampage V Extreme) and it does not have the RSR setting listed.

Also, in my stock Clevo bios 1.05.01LS2 / EC 1.05.04 from Clevo does not list this setting either, but I'm able to bench at 4.9 ghz with it. And I'm 100 percent sure it's disabled in the bios by default.

 

I did some searching and found out that some Gigabyte and ASRock desktop boards have an option in the bios to enable or disable RSR. Asus desktop boards disable RSR in the background since no one in their right mind is going to enable throttling when they are overclocking.

The chance of finding the magic bit that controls this is not good but I am willing to have a look if anyone can help out. I need a person to boot up with RSR disabled in the bios and then do a register dump. After that, reboot, go into the bios and enable RSR, boot up and do another register dump. You can use my RegReport program to dump your register contents to a .txt file. Send me or attach the 2 reg report text files and I will start searching.

RegReport
https://www.sendspace.com/file/isttre

 

Doing it now.

Edit: Ok, i'm still laughing over here!

Edit: files sent.

 

Not much on TV this time of year so it is nice to have some numbers to look at.
Thanks @Johnksss@iBUYPOWER for helping out.

 

No problem!! I can give more if need be. If you are bored. LOL

 

Well despite my RSR problems I just managed the #1 slot on firestrike for the combination of a single 1070(notebook) with a 6700k, so I guess I'm not doing too bad

 

@unclewebb - trying to understand speedshift with throttlestop 8.40 and kabylake (win10)

when it's enabled from the tpl window what setting should be put on the min and max for
-max performance
-max power savings?




and then there's speedshift epp, can be triggered by the different profles, do we need to have throttlestop "turn on" to use this feature?

also a question on how the setting works because you can set a value next to it, what should be put on the box for maximum performance and maximum power savings?




from this post I understand that: http://forum.notebookreview.com/threads/dell-xps-15-9550-skylake-speed-shift.796891/

epp value lower = more performance
epp value higher = more energy saving but cpu is slowed down
-but needs to "turn on" throttlestop?


on tpl window you can adjust the min value and max value, this represents the min and max multiplier of the cpu?



how do we know in the os that speedshift has been enabled?



correct me if I'm wrong, just trying to learn the software here


also how do I make an information dump for the cpu? with kaby lake I guess you need more data from testers

 

255 across the board.

 

@Mobius 1 - The Speed Shift Min - Max settings in the TPL window controls the multiplier. A setting of 255 will not give you the 255 multiplier but it should give you the maximum multiplier. A setting of 1 will not give you the 1 multiplier since that does not exist but it will give you the minimum multiplier which is usually 8 for many Intel CPUs. Some of the U CPUs use a lower minimum multiplier like 4 or 5. If your maximum multiplier is 40, it shouldn't make any difference if you set the Max to any value between 40 and 255.

I like Speed Shift because it gives users more control over their CPUs compared to previous generation control schemes. It gives a user so much control that I think some manufacturers are afraid to enable Speed Shift in the bios. Why? Because it can also be used to disable some of the popular throttling schemes that manufacturers have dreamed up over the years.

If you do not like your CPU dropping down to the 8 multi when lightly loaded, it would be easy to set the MIn to 16 or whatever multiplier you like. If you think a high multi is a bad thing during certain situations, you could easily use the Max setting to cap the maximum multiplier. Warning, @Johnksss@iBUYPOWER won't be your friend anymore if you start using ThrottleStop to reduce maximum performance.

The EPP part of Speed Shift lets you adjust how fast your CPU goes from idle speed to full speed.

EPP = 0 should get you max speed all the time. This setting should be similar to the Windows High Performance profile.

EPP = 128 is the default Intel recommended setting and this will get your CPU performing similar to the Windows Balanced profile where the CPU slows down when lightly loaded and gets back up to full speed when loaded.

EPP = 255 will put your CPU into slow and sluggish mode where it will take a significant and sustained load to get your multiplier to increase at all. It probably won't get anywhere near the maximum multiplier, even when fully loaded with Prime95. Keep in mind that a slow CPU might not be the way to get maximum battery run time so do some testing before jumping to the conclusion that slow and inefficient is a good thing. Software based power meters like the data TS shows you is not accurate when a CPU is lightly loaded so don't use that info to make any decisions about efficiency.

I have done zero hands on Kaby Lake - Speed Shift testing. If you notice anything totally different than above, post some pics. ThrottleStop lets you enable Speed Shift in any Windows operating system so it might be useful if someone wants to run Windows 7 on a new laptop.

 

I'll play around with the settings.

So from what you say, that the min should be set to whatever the cpu is observed to be lowest frequency, and the max is 255 to be safe?

I'm trying to play with epp, you're right though, 0 feels normal but 255 makes the laptop very slow. 128 has minimal effect that I can observe from this 7500U.


I don't think I noticed something odd yet, but I can send you data for future testing purposes.



Also how do I know that speedshift is active/enabled by ts?

 

I have max performance goal so on my i5 6300HQ. In ThrottleStop 8.40 I set SpeedShift as follows:

MAIN SCREEN
*"Speed Shift-EPP"
- check box
- Type "0" in the box

TPL SCREEN
*Next to the the word "Speed Shift", these figures came out as default on ThrottleStop
- Min ="1"
- Max ="32" (becuase max turbo speed of my i5 is 3.2Ghz-yours may be different)

*"Enable Speed Shift when ThrottleStop starts"
- check box

A few posts up, Uncle Webb noted that in the main screen of ThrottleStop, if you see the letters "SST" in green then SpeedShift is enabled with the EPP in the neighboring box.

Finally, keep in mind the EPP curve from 0 to 255 is not linear. Someone posted his results android compiling in that 9550 SpeedShift thread so you can get an idea of that.

 

Many people are familiar with checking HWiNFO to see if Speed Shift is enabled, so I copied his great idea. If ThrottleStop reports SST in green, Speed Shift is definitely enabled.



When using HWiNFO to monitor Speed Shift, you will need to restart HWiNFO after you resume from stand by so HWiNFO updates this properly. Once SST is enabled, it remains enabled until you either reboot or resume from sleep. Windows 10 should handle this for you but some motherboard manufacturers have decided not to use Speed Shift.

There is no unsafe setting so don't be afraid to play around with this. You can set Speed Shift however you like. Just keep an eye on the reported multiplier to see if the CPU is running how you would like it to run.

 

what's weird is that with 7500U turbo boost (can consume above 15w) the chip will slightly throttle (not exact 3.5ghz, but slightly below that) - but throttlestop reports that there's no TDP throttling - I can't really be sure of what's causing this








something else I notice is that the PL1 changes dynamically with this particular laptop (Envy X360 13" 7th Gen ULV), I am not sure what's causing it but I can provide an info dump to see if you can identify the problem

I have seen the PL1 drop as low as 9watts, hovering between 10-12, very weird

This if false advertising, under heavy occt load with sufficient cooling (under 70 with cooling pad) the cpu will not sustain stock clocks!

@Mr. Fox @D2 Ultima @Johnksss@iBUYPOWER @jaybee83 @ole!!! @hmscott @Papusan @Phoenix @TomJGX @Ashtrix

maybe @Prema can apply a bit of magic to this poor laptop and see if the PL1 dynamic throttling can be fixed?

@triturbo I will get pictures of the heatsink and fan up close for you later

 

Sir, sir. That is not dynamic throttling. That is dynamic power usage to provide the best experience with all the power a user could ever need! That is the future! We should all be throwing our hands up in joy willing to throw more money at HP and BGAtel for such a wonderful CPU, killing its used power against our will for our own good!

Now if you'll excuse me, I have to mop up all the sarcasm I spilled over my keyboard. And the floor. And the neighbourhood. Basically, I caused a sarcasm flood in my country, and might be arrested soon. I WENT OUT FIGHTING FRIENDS.

 

The max. Turbo Boost Clock is only reached with single core loads, meaning no load on the other core(s) at all. Even if you just run a Benchmark on one thread, the OS and other processes eventually will wake up the other core from time to time, which is why you never will be able to constantly maintain the full turbo boost frequency in a real world scenario.
You can check the Multicore Turbo Boost Limits in ThrottleStop: FIVR -> Lower right corner (Turbo Ratio Limits)

What did you use to monitor the PL1 threshold?
If it's really changing on it's own, i reckon it's some special feature in the BIOS/UEFI. That usually means you can't do much about it, as long as the vendor hasn't added a Checkbox to the setup menu - did you check your BIOS settings, if there's some option related to this?

---

OT @TopTension (sorry):

Spoiler: OT

Me neither, but unfortunately I'm not privileged enough to send a PM yet (that's why I could find the stupid PM-Button in the first place, damn).
I were quite busy during the holidays, but I'd be really interested to try it out and maybe adapt and improve it a little. I'm not a real windows dev, but I think I'll at least get it working on my machine
Do you think you could send me a message (hopefully I can reply), because otherwise I'll have to wait until I've manged to write at least an other 3 Posts (as far as I know now)...?
Thanks in advance!

 

hwinfo


I have to override it everytime using xtu to keep the tdp 15w


this is really false advertising by hp

 

Mobius -

The Intel XTU program looks awesome but has some bugs with recent Intel chips. i5 and i7 Skylake users (9550) including myself reported power throttling caused by XTU. Similar to what you are suffering. There is a long thread here at NBR if you want more info, but just try the following for now:

**** Please completely remove Intel XTU program from your system and try testing***

***Regardless, you should not have XTU and programs like ThrottleStop running as they conflict and confuse each other***

 

I will try removing the XTU using revo uninstaller and see if the problem persists.


Also I try triple check the CPU boost multiplier is 35x for both cores, I can't adjust up or down.

 

http://ark.intel.com/products/95451/Intel-Core-i7-7500U-Processor-4M-Cache-up-to-3_50-GHz-?q=7500u

The 7500U has 3 different modes. Open up the ThrottleStop TPL window to have a look. The default TDP is 15 Watts but there is a Configurable TDP-up level of 25 Watts and a Configurable TDP-down level of 7.5 Watts. When you buy a U CPU, that's what you get. 3 CPUs in one. The U can change significantly depending on which TDP Level it is operating in.

Don't run HWiNFO when using ThrottleStop - Limit Reasons. Use Limit Reasons by itself and it should show you what is causing the throttling. This is going to vary depending on how much load you are trying to run and what TDP Level it is in. The different TDP Levels also have different default multipliers associated with each level. When the CPU is stuck in one of these levels, the multiplier will drop down and the power consumption when throttling will not be exactly 15 W or 7.5 W.

It is not false advertising. Just Intel and computer manufacturers preying on uninformed consumers. Maybe Trump can clean things up and make the computer industry great again. Try sending him an email. I heard that he doesn't check his email very often though.

It looks like Intel has changed this with Kaby Lake.

http://www.cpu-world.com/CPUs/Core_i7/Intel-Core i7 i7-7500U.html

The 7500U has the same 35 maximum multiplier whether 1 or 2 cores are active. Looks like Intel is using strictly power consumption to control these CPUs.

Almost forgot, this is a U CPU not a K CPU so the turbo multis are probably locked and not adjustable.

 

First off, thank you, @unclewebb, for this amazing program and all your hard work!

I currently have ThrottleStop set up with two profiles, the first being maxed out performance mode and the second with Turbo Boost disabled to reduce heat and temperature. I enabled the DTS Alarm feature in the options, and it works as intended. As other people have mentioned, the two profiles switch constantly, causing rapid fluctuation between clock speed and temperature.

A type of timer would be ideal, so I did some digging and found this profile delay feature you added here:

It almost works perfectly. However, whenever I enable a delay to profile 2, the one the alarm is set to switch to due to high temp, the switching from profile 1 to 2 occurs with the countdown timer, but the switch back from profile 2 to 1 never happens. It just stays stuck on profile 2.

For example, the following only switch from profile 1 to 2, but stay on 2:

Code:

DelayProfile1=2
ProfileDelayDuration=3

-or-

DelayProfile1=2
DelayProfile2=1
ProfileDelayDuration=3

Interestingly, the following is the only way to cause the switching back and forth to work properly, but then I can't set a delay from profile 1.

Code:

DelayProfile2=1
ProfileDelayDuration=3

Note that, ignoring the alarm, the delay countdown works as expected in the GUI when manually clicking a profile's radio button for any of the previous configurations.

Since the profile delay looks to be a relatively hidden, and possibly underused, feature, I was just checking to see if I perhaps have something misconfigured or if there is some sort of conflict between the alarm and delay in the program.

Any help or guidance would be much appreciated!

 

@hmb39 - I can honestly say that I do not remember writing a single line of DelayProfile code. It was so long ago that I have no idea if it originally worked correctly or not. Besides nando4, I don't think anyone else was interested in this stuff. It was definitely nothing I ever remember using.

If I had a laptop with a Core i CPU, I would either try to improve the cooling so you can run at full speed without overheating or I would try setting a lower TDP power limit so the CPU consumes less power, produces less heat and doesn't overwhelm the inadequate cooling solution.

I will try to have a look at these old features early next week to see if they still work.

 

hmm, I don't think I have limit reason when I downloaded throttlestop? is it a separate feature on throttlestop that I missed?

 

@Mobius 1 - I think you need to go back to ThrottleStop school to learn about all of its wonderful features.

 

greyed out