The ThrottleStop Guide

I get the exact opposite... sometimes. Turbo may be disabled if I set the value higher than 23. I don't think that is the intended purpose of the function.
The Set Multiplier works for me just fine. I've actually had to use the Set Multiplier to override the Non Turbo Ratio glitch I got (I never locked the NTR before).
Until now, locking NTR seems to work... most of the time. But now I cannot unlock it (to see if an unlocked NTR was the problem).

 

The 4700MQ has a default multiplier of 24. If I set the Non Turbo Ratio to 24, Set Multiplier offers full control of the multiplier for any value of 24 or less but I have no control of any of the Turbo multipliers higher than 24.

A Non Turbo Ratio setting of 0 gives me full control of all of the multipliers from 8 to 34.

If you lock the NTR, you will need to reboot or do a Stand By - resume cycle to unlock this register. Make sure that ThrottleStop is not still set to Lock or it will lock this register over and over again.

 

@unclewebb it seems that when I close ThrottleStop, the VCCIN values get messed up, or swapped around. Both Normal/Game profiles are set to 2.1250v and Battery is set to 1.7305 (though pretty low, it seems to work fine. Turbo mode's disabled and it's locked to 2.0GHz, so this might help). If I close ThrottleStop and reopen it, sometimes Normal/Game VCCIN is set to 1.7305 and Battery is set to 2.1250v. Sometimes, it shows "Default".

I know VCCIN is working because it directly affects the time I can run on a Turbo setting, by tricking the Package Power value. Also, because setting it too low crashes the computer
But whenever I completely close ThrottleStop, it messes up the settings. The "Offset Voltage" is kept like always, but VCCIN gets modified every time.
You also said to turn OFF "Unlock Adjustable Voltage". Didn't you mean to turn ON? When I turn it off, the last VCCIN value is kept through all profiles, and it's also kept if I close ThrottleStop. This is also something to consider.

Is it possible that my C state timer got messed up somehow, and it's taking a extremely long time to trigger the C6 state? I don't remember messing up with it, but who knows.
The Y50's 4700HQ does enter C6 state, as @ATG has one like mine, and it works on his. I'll have a look at his powercfg output and compare with mine, to see if there's any hidden option turned off.

--------------------

I've compared mine and ATG's power profiles and they're identical.. It ain't a windows setting. There must be something else.

 

If I set it to 24 where my default is 23, it's normal for Turbo to be disabled? This is what I'm asking because this doesn't seem intended.
I've retested this and can confirm that on my laptop, a non-lock NTR set above the default will cause Turbo to be disabled. Outside of that particular setting and in low battery health, Turbo remains enabled for my laptop.

 

@Game7a1 - Your 5th Gen U CPU works differently than my 4th Gen non U CPU. ThrottleStop works the same on either CPU so I have to assume that either Intel made a change between generations or the U CPUs work differently than the non U CPUs. Maybe a bit of both. The only important question is does this new Non Turbo Ratio setting give you some control over your CPU? This new setting might not be of any use for your particular laptop but maybe somewhere else, a manufacturer has set this register differently and now a user can go in and adjust this to whatever they like as long as it is not locked. Not every ThrottleStop feature is going to be useful for every user or every laptop model. I try to add features that might be useful to someone and then sit back and wait for some feedback.

@GTMoraes - Sorry for the confusion. When I said to check off Unlock Adjustable Voltage, I was thinking about a list where you "check off" items. I didn't mean to leave this option in the off state. I did an edit and removed the word off to avoid confusion. Thanks.

It has been an interesting day. I went through the data you guys posted and I found a register that seems to have a direct effect on what Package C States a CPU uses. The Y50 uses a slightly different value than my Y510P so I decided to add a new option to the ThrottleStop C States window so you can play around with this setting to see if that changes the percentage of time your CPU spends in various Package C States. This new option has given me full control of my 4700MQ so I am hoping that it might help you out @GTMoraes.

The Intel docs call this the C7 Interrupt Response Limit. A setting of 6 gives me about 82% in Package C6 and the reported power consumption is showing 1.4 Watts. I need to do some more testing because the power consumption data provided by Intel Core i CPUs is not 100% accurate, especially when a CPU is idle but it is interesting all the same.



Next up, I adjusted this down to 5. Same load on the CPU as the previous screenshot but this setting completely disabled my C6 Package C State and reported power consumption went up a little too.

http://i.imgur.com/xU0UJeI.png

A setting of 4 wiped out my C3 Package C State so now my CPU was spending most of its idle time in the C2 Package C State. Power consumption was up again.

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

Anyone seeing a pattern here?

A setting of 2 and my 4700MQ no longer uses any of the Package C States when idle. Reported idle CPU power consumption is way up to 7.7 Watts. I don't believe the accuracy of the power consumption data but the package core temperature is also up so there is definitely something going on.

http://i.imgur.com/2VzvFRm.png

Even though Lenovo locked the Package C State Limit register in the bios, I now have a way to adjust what Package C States my CPU will use.

This new feature is mostly hooked up and working so a beta 4 is almost ready. Maybe this new feature will improve battery runtime. Maybe significantly.

 

Locking the NTR works as it should on my laptop; not locking it may not work properly under a specific circumstance (one that can be made carelessly).
I understand that this feature may be of importance to someone else, but I thought it would be good to point out that there may be a bug.

 

I have a suggestion @unclewebb for the next iteration of TS8. Could you have it report the VCCIN values in the FIVR menu when it's set to default? I haven't touched VCCIN yet, but in the window at the top right where it shows cache ratio and various voltages, VCCIN simply returns "default". It would be nice to know what my "default" actually is; the slider for VCCIN already has a "default" setting there.

 

Oh boy, oh boy! Can't wait for the C7 IRL slider! I have a good feeling that this might indeed work for my case, and if these Power readings are anywhere near the real deal, it'll be awesome for increasing battery life and temps when unplugged for everyone!

Thanks for your hard work. No, seriously. Thank you.

Oooh, I got it. Didn't undestand it in that way, as you can check it "off" -- in the sense of turning off. But now I got what you've tried to say.

The boxes are checked, and "Unlock Adjustable Voltage" is enabled for every profile. Undervolt options work and are saved correctly. I also checked for saving voltages when OK is pressed, but the profiles VCCIN still changes when ThrottleStop is closed and then reopened (or if the machine is rebooted).
If I change the VCCIN and then uncheck "Unlock Adjustable Voltage", the new VCCIN is kept for every profile, even if ThrottleStop is closed and reopened.

My FIVR window is set up like this:
http://i.imgur.com/71B5KhN.png

The VCCIN on the .ini is set up like this:
http://i.imgur.com/TznjnFy.png

After I close ThrottleStop, open it up again, click on the FIVR button and then press OK, it is set up as that, on the .ini:
http://i.imgur.com/9rrJYSS.png

And if I repeat the same procedure again:
http://i.imgur.com/codocA4.png

 

@GTMoraes - Thanks for your detailed explanation. I do a lot of copy and paste kind of programming where I will write a small chunk of code and then copy and paste the same code for each profile. Sometimes I forget to edit the stuff I just pasted to make sure it works on the correct profile or saves and reads the data to and from the correct INI key. Your testing will help me track down this bug.

Unfortunately this is not possible. The default value in the VCCIN register is zero and zero doesn't translate to any particular VCCIN value. Intel might know some secret but they are not sharing. It is not until you start adjusting VCCIN that you can have a look in the monitoring table to see what value you are actually getting. If the VCCIN register has a zero in it, ThrottleStop simply reports that in the table as Default.

To try and find out what the Default value really is, I ran the TS Bench, watched the power consumption at the Default value then set VCCIN to a fixed value and tried to get about the same power consumption value to display. When using a high VCCIN, power consumption dropped down to less than what I normally see. I kept lowering VCCIN until power consumption more or less matched and at this point, VCCIN was about 1.80 V. Not an exact science and I didn't spend too much time doing this but for my 4700MQ, somewhere around 1.80 V seems likely.

 

Well that sucks.

 

Fine job on the new version, unclewebb

Just passing by to congratulate you once again and to report that 8.00 beta 3 is working flawlessly on a Z97 system with an i7-4790K.

On the whole C State matter, is there any particular reason why I'm not finding any actual changes in Core C State percentages between C7 and C7s?

 

Add another one of horrifying crashes. This time around, I'm still shaking. So much that I feel my fingers numb. And my heart is pounding so strong, that if I didn't made my checkup few weeks ago, I'd think I'd fall stiff from the chair at any minute. Damn

Here's the story:

tl;dr: I went over the edge while messing with the voltages. Computer suddenly turned off and refused to turn on. Scared the crap out of me. Seems fine now
Also, I had some flip-flopping frequencies behaviour while stressing the system under this very low voltage setting. I'm still unsure whether if it's due to processor instability or some low level protection.

Remind me to not go way over the edge next time.

 

Yeah let me give you some nice advice:

- If it doesn't have a "watchdog" to revert from crashes like XTU does, and it affects overclocking? Don't set it to run on boot.
- If your machine shut off and won't turn on after fiddling with voltages, unplug everything and remove CMOS battery for a few seconds, then plug the CMOS battery, regular battery and PSU back in and try turning it on again.
- Black screen "poof" off states are a result of too low voltage SOMEWHERE on the system. Not enough for the processor to actually run correctly, but enough that Windows doesn't think there's a problem (WHEA_UNCORRECTABLE_ERROR BSODs are of course Windows' doing). I used to get this under sustained heavy load with my CPU when I set it to 3.9GHz and used default voltage. 3.8GHz was fine and never tripped though. In other words, they're normal, and you shouldn't be scared SPECIFICALLY because of that (though I do understand you were scared that it wouldn't turn back on; that's legitimate).

 

I thought that, differently what happens on desktop PCs, laptop's CMOS batteries were just for keeping the clock ticking.
At least, if I set a boot password, and then remove the CMOS battery (and the regular battery ofc), it won't reset the password. Infact, it won't reset any setting at all. That crossed my mind at the time, and helped scare me slightly more. I didn't try holding the power button with all batteries disconnected, though.
I'm more used to messing with desktops. Screwing things up, you'd just remove the CMOS battery and tis done. Newer models don't even need to, you just let it detect your mistake, and it'll boot up with an "Overclock error!" screen. Marvellous.

I've had a few WHEA errors (even a orange BSoD once. Pretty cool), so I was kinda used to it. Knew the cause, and it didn't happen instantly after boot, so was no biggie. Sudden blackout and having a hard time to turn on? Never. I really thought I had killed it.

Funny thing is, it managed to hold its own when I was stress testing it with the low voltages for about 20 minutes. It died out of sudden during the beginning of a stress test. Only possible error was the flip-flopping of the frequencies from max to min, which I was guessing it was some low level limiter (and because of that, I also thought I had tricked the power consumption way too much, and burned out some component. That was the main scare).

I guess I won't adventure too far from my safe values anymore. Perhaps I'll decrease a notch every now and then, to find the sweetest spot. But for this week, I'm settling the voltages as it is lol

Thanks for the tips! It's good to know all I've went through was expected. I'm a novice when dealing with laptops.

 

Guys the link for ThrottleStop 8 is not working, why is the file deleted? please send me the link, Thanks in advance

 

Please man send me the link for throttlestop download

 

Did you read the creator's signature?

https://www.sendspace.com/file/tmc96z

 

sorry man I didn't , but thankyou very much for help, I am actually using the alienware 15 with 7 4710hq which I want to overclock without any throttling, choosing this over intel xtu is a good idea right? one more question, my intel thermal framework used to reset the value of max turbo power long sometimes even lower than the stock, will I experience anything like this on this software too?

 

I suggest you make yourself a cup of coffee and read a few pages back.

But yeah, I prefer ThrottleStop over XTU anytime. And most likely you'll be able to keep your turbo clocks, unless it's temp throttling.

 

Okay man you are right but please make my work a bit easier, I am already very much troubled by throttles, and fixing issues, please just answer my questions

Just let me know if I need to delete intel thermal framework?

 

Try condensing your questions in one post, please.

I've never dealt with this "Intel Thermal Framework". Never seen it, didn't knew it even existed. Pehraps someone else over here might help you.
Or the answer might be already in this thread. Use the search function

 

Removing the CMOS battery resets lotsss of things in my BIOS. If it does nothing to yours... I don't know. Screw Lenovo? *shrugs*

I've never had a hard time to turn it on like that, but sudden blackout I have had.

*meows*

 

You will never be able to overclock that CPU without throttling. You need to sell your Alienware and buy a desktop-CPU-using Clevo laptop if you wish for overclocking on CPU without throttling.

Generally so, unless you need features from XTU that Throttlestop does not offer such as XMP adjustments for RAM.

What in the living hell is the Intel Thermal Framework and how does one delete it?

 

Hi there,
First of all, really nice program Unclewebb.
- When I click on save in the main window, my harddrive start to make strange noises, like if it was going to sleep every 3-5 seconds. Restarting the machine resolves the problem.
- When I only use one core, all cores go to the maximum frequency, is it normal (sorry for newbie question)? It leads to a package consumption of 6.8W compared to 7.5 will all 4 cores (2 real cores with HT)
- Is it possible to set a maximum coefficient? I only found a way to fix a coefficient which leads to higher consummation in idle state.
- Does someone know a low power test for stability? My Laptop is very stable with -0.12v on cores, CG and cache while running Prime95, but crash and restart when doing internet things. I went back to -0.1, but I don't want to have corrupted files just because some minor errors happen without leading to a crash. I tried Prime95 with one thread, but it's also very stable at very low voltage. I know it's strange but it's real, so any help appreciated
Regards

 

Kinda sucks indeed. Makes me wonder such extreme scenarios. While others (having a password) it sounds pretty nice. I'm unsure.
But well, it seems that changes made on ThrottleStop aren't kept after the computer reboots, so I don't need to reset the CMOS.

But oh boy, I'm still scared of that sudden blackout. I can take a garbled blue screen and a reboot any day, but this sudden blackout surely shook me up.
Livin' and learnin'

 

1 - Like a "heartbeat" noise? Had this back when I've used HDD's, but I thought it was related to the HDD APM or some kind of this stuff. I don't think TS messes with these.
2 - Yeah, sounds normal. I've never seen the cores working with different speeds for a long time (+1s), though.
3 - Coefficient of what?
4 - Best low power test for stability I know of is to stress test it for a couple of minutes, then use it as usual. You can also use RoG RealBench to fully stress test your computer, and using real world scenarios like decoding stuff.
4a - My laptop seems to be able to run at a very aggressive undervolt when it's set to a low frequency. Sounds fair. Perhaps yours do too.

 

1 - I find it strange, but it's reproducible. Every time I click on save, this sound comes.
2 - Ok, if nobody says the contrary. It's a bit of a power loss, but ok.
3 - The frequency multiplier. It's between 8 and 19 by default. For some profiles, I would like to have it from 8 to 12 for example (to get low consumption in idle, some perf boost in browsing internet but not full power drain.
4 - It's the opposite for me. With high under-volting, it's stable in stress test (Prime95 on 4 threads) but it crashes and reboot the laptop in Chromium/Office or other low power programs...

 

1 - Then it's something for @unclewebb to look for. I can confirm I used to hear this noise back when I used an HDD (if it's the very similar sound like a heartbeat), but I can't confirm if it's directly related to ThrottleStop.
2 - I guess it's because the processor can't work with different speeds on certain areas. That's just an guess, though.
3 - Oh I get it. Well, you can lock it to a certain multiplier, and enable the Power Saver button. You can also define how aggressive the power saving feature is, on the Options menu, under Power Saver C0%. Mine's set to 50, so it's a bit more aggressive in keeping the frequencies low. Does yours shows some multipliers with an "T" next to it? If so, check "Disable Turbo", so it doesn't go on Turbo, and over the frequency you've set.
4 - That's expected, it's why I said to use it as usual. When you're running a stress test, you have a steady frequency and a stable processor voltage. If your undervolt is unstable in that scenario, it is completely unstable in any other. In normal usage scenarios, your processor goes back and forth in frequencies and voltages, adapting to your needs, and reducing power on lack of demand. This is the stage that some undervolts fail. When I said that "my laptop runs an aggressive undervolt when set to a low frequency", I was talking about 8-20x, heh. It, as default, goes from 8-24x, then 25T-36T. That's a rather large range for the undervolt to work on, and the voltage is naturally set higher by Intel to keep this power hungry processor working. On your case, it might not make much difference using a "lower frequency".
Also, every processor is unique. I know mine has some hard time around -55mV, but I've seen others working fine at -60mV and -70mV. Your mileage may vary

 

Ok, so 2,3 and 4 are cleared, thank you For the strange repetitive and reproducible noise, is there a bug report forum? I have an I3 4030U (exact computer model is Medion akoya-s6213t-md-98845).

 

AFAIK it's right here. Unclewebb is pretty active, and he might say something about this noise sooner or later.

 

Is it safe to change the voltage of analog or digital I/O or can it lead to data corruption? Does it really bring power saving as much as the cache/CPU/GPU part?
What is the system agent voltage?

 

from my experience:
CPU/Cache = Undervolting gives many gains
GPU = Touchy to undervolt. Might lead to unexpected instabilities (as the IGP isn't used as frequently as the CPU)

All the others: Didn't notice anything different in the Package Power. I've used values as low as -80mV and nothing happened. I just set it to zero and didn't mess with it anymore.
It'd be interesting to know what do they actually affect, and if the undervolt applies successfully.

 

@TomJGX Tested Wprime 1024 stress test for you sir... Quite equal previous performance with 4.3ghz. 167.46 (Throttlestop 8.0 beta3) now vs 167,657 in the previous stress test and old beta2. Within a margin of 0.1% but with better score. LOL
19 months old Liquid ultra give still good temperatures in Wprime 1024 stress tests
New. http://hwbot.org/submission/3005419...e_i7_4930mx_2min_47sec_460ms?recalculate=true
Old. http://hwbot.org/submission/2849851_papusan_wprime___1024m_core_i7_4930mx_2min_47sec_657ms

@D2 Ultima was wondering about Alienware CPU throttling... ? Has he forgotten that my AW17 does not have the useless new modern Intel BGA processor aka Hq?

 

I was talking about the HQ chips, not about your AW17

 

LOL. @TomJGX is on the wrong track .

 

yeah, one of dem ...

"useless new models"

 

Hey folks, I'm still trying to guess what's causing my Package C state to not fall into C6.. I'm now guessing it might be the IGP that's holding it in a powered state. The Intel ARK says my IGP is supposed to idle at 400MHz, but it idles at 599MHz, with a brief time at 200MHz, exactly at the moment when I remove the AC cable.

Could you guys check at which frequency is your IGP running right now (and if it's different from my IGP, a HD4600, please inform the model), and if your processor is entering Package C6 state? You can use GPU-Z to look for this information. Thanks

 

Same IGP, same values and behavior as your.

Edit: Maybe its something in your TS config, if thats even possible...Have you considered it? As I mentioned the screenshot I've posted was made with TS running stock, apart from the config changes you can see on the screenshot itself.

 

Oh well.. back to the theories board.
Yeah, I've completely wiped my SSD and reinstalled a fresh windows, then extracted ThrottleStop from the zip and ran it. It even showed me the initial warning and stuff. C6 was still dead.

I have to start with crazy theories... like that when I updated the BIOS to 3.03, I had ThrottleStop running, so it overwritten some stock settings on BIOS, thus disabling deep C6/C7 support.. or it's the damn aliens who's doing this.
The latter sounds more plausible.
For the former, I have to overcome my personal fear of flashing BIOSes.

 

A shot from CPU-Z though I realize its probably useless but just in case.

 

I'm too brain dead and overloaded to think.. Yeahhhh, so not my fault :

 

Hey guys i know that i am too new to say anything like this on this thread, but i really have been using throttlestop for a long time now, I just want to say that it is a software of much more use that just stopping throttle, it is much better than Intel XTU and needs to be no. one overclocking utility for CPUs, and for that i strongly believe that it should a have name change to something like CPU Overclock Utility so that people download it for overclocking too, i also think that it should have some kind of safety feature which can prevent people from going off limits if the feature is turned on.

 

What would be "off limits"?

Thanks for the shot. Mine's exactly the same, no more nor less. I'm about to request a BIOS advanced menu unlock, and I'll try flashing it. It's my final hope.
Still have to overcome the fear of flashing BIOSes, though.

 

I've flashed and unlocked my BIOS. Feels like I've overcome a great childhood fear (srsly, I feared flashing BIOSes since, idk, 10?)
The only lasting fear is to change any setting that I shouldn't and brick the laptop. Given the CMOS isn't reseted when the CMOS battery is removed, this is an acceptable fear.

I know this is -WAY- out of the scope of ThrottleStop, but I'd like you guys could check the C-States options I now have, and suggest any modification, if known.
http://imgur.com/a/Q4sh7

I'll try keeping on the ThrottleStop subject for now on... no, wait, just one more thing: It seems I can disable the TDP throttling limitation. Should I, or does it risk the motherboard circuitry?

Ok, subjects will be strictly ThrottleStop related from now on

-----

btw, no C6 state either, with these settings.

 

ThrottleStop 8.00 beta 4
https://www.sendspace.com/file/78dnne

New Features:
- fixed Skylake C0% reporting
- reduced max VCCIN slider from 2.50 V to 2.30 V
- fixed VCCIN profiles not being saved correctly.
- fixed Haswell reporting C8 / C9
- added C7 Interrupt Response Limit adjustment.

Thanks @GTMoraes for finding the bug where VCCIN values were not being saved to the correct profile. That's fixed. Skylake C0% might be working correctly now but I need someone with a 6th Gen CPU to post some screenshots at half load and at full load.

For the 4th Gen and up CPUs, I added an interesting adjustment which lets a person tweak the C7 Interrupt Response Limit. I am still not sure whether adjusting C7 IRL gives a real benefit. A low value for this seems to disable the C2, C3 and C6 Package C States on my 4700MQ which might make the C State haters happy.

Here is an example with C7 IRL set to 11. It shows a very high percent of time in the Package C6 C State and a very low power consumption number of 1.2 Watts.

http://i.imgur.com/FhdRds4.png

In the second picture with C7 IRL adjusted to 3, same CPU load (C0%) but now all of the Package C States are disabled and power consumption is way up at 7.7 Watts.

http://i.imgur.com/mEwfLx7.png

If power consumption monitoring in the Intel Core i CPUs was 100% accurate, that would be significant. I am not sure about this new feature so I decided to include it as a possible new feature to see if anyone finds a use for it. Move the IRL slider back and forth, hit Apply and see what you get.

I used the Windows Performance Monitor - Battery Discharge data to try to see if this is really reducing power consumption but I am not sure how accurate that data is.

 

Hey, it seems that HWiNFO64 can report the current power consumption by measuring the battery remaining Wh drain rate (which seems precise).
Although it reports the value at a relatively slow rate (it's not an instant measure), it seems to pick even increase in consumption due to the display brightness
http://i.imgur.com/sIJY86D.png

After you remove the AC cable, it takes a while before it goes from the charging values (positive) to the discharging values (negative), and another while until it stabilizes in the current discharge rate. It's good to maintain the computer completely still while it's reaching the steady rate. Idk the Battery Discharge data from Windows Performance Monitor, so this might be redundant. But it's the tip I can give =)

Thanks for the release!!

 

HWiNFO, Battery Bar which I use and the Windows Performance Monitor, as far as I know, are all getting their data from the same Windows function. I have never compared this data to measured power consumption but it might be reasonably accurate if a laptop has had a chance to stabilize.

The power consumption data displayed by ThrottleStop, CPUID HWMonitor, etc. is calculated by the CPU internally so the CPU can decide whether it should be using full Turbo Boost or if it should be throttling Turbo Boost. ThrottleStop is a tool to monitor throttling so I thought it would be a good idea to include this data to help users understand why their CPU is throttling.

The problem is that people trust this data too much. When a CPU is idle and the cores are spending almost all of their time in the low power C7 C State, this power consumption data is not accurate. It is not the same as measured power consumption.

Not sure what that is all about. Some laptops use cheap capacitors that squeal like a pig during some situations. Using ThrottleStop to change the CPU speed, C States or voltage might be a problem for your laptop. ThrottleStop does not do anything to your hard drive.

Read the above. Software power consumption reporting is not 100% accurate. Don't use this data to make comparisons of small changes in power consumption.

When a Haswell CPU core is idle, it should be in the C7 C State where it is disconnected from the voltage rail. In this state, a high or low multiplier does not make any difference. It is like a bear hibernating in his cave. It is oblivious to what the rest of the CPU package is doing. The higher power consumption you are seeing is based on data that is not 100% accurate.

An excellent paper called Power Optimization - A Reality Check helped me to understand that a high multiplier is not as bad as you think it is. It might actually be a benefit.

http://www.cs.berkeley.edu/~krioukov/realityCheck.pdf

I have the same problem. An under volt setting that is 100% Prime stable might cause a BSOD or crash when the CPU is lightly loaded. The problem is that the C States are already lowering the voltage. The under volt combined with the C States causes the CPU voltage to drop too low so you quickly end up with a light load crash. I think there is a way to set a minimum voltage when using the C States to prevent the CPU from dropping the voltage too much. I plan to look into this further because I think it would really help your situation. I would love to be able to run my 4700MQ all the time with a -100 mV or -120 mV under volt but when lightly loaded, this is not yet possible. Anything beyond -40 mV and I have the same problems with crashes when lightly loaded. Maybe using my new C7 ITL trick to disable the Package C States might help out.

Thanks for your feedback.

 

Oh, I see. Thought it was getting data from another source or something. I thought it was fairly accurate when charging, and displays the "Rapid charge" mode accordingly. But yeah, even by letting it stabilize, it fluctuates a little bit.

And yeah, I've been having this "issue" of trusting that number way too much, heh. Gotta learn to not consider it that much.
Maybe if you change to "Estimated package power"? heheh

This is really interesting. I've been thinking about that when I've read someone commenting (I guess it was you, again lol) about how it's better to have a processor running fast and finishing its job quickly, than a processor running in a "economy mode", but taking longer at a high usage state.
I've set my processor to stick at 1.5GHz when on battery, but I didn't really notice any improvement in battery. But I did indeed notice a impact in overall performance. I also used HWiNFO to measure the average frequency my processor was using during my classes. At a maximum of 1.5GHz, it had an avg. of 1.4GHz, and at an maximum of 3.3GHz, it had an avg. of 1.6GHz. Battery duration remained about the same (was crying for a charger at the last minutes of the class), but performance was severely crippled when at 1.5GHz. I guess I also felt it a bit hotter, too.

I realized a very informal test yesterday, comparing avg frequency, temperatures and estimated power consumption over a small battery percentage span, at a maximum of 1.5GHz setting and a maximum of 2.4GHz, under "normal usage" (facebook, checking my tracking codes periodically, checking the mail, a bit of youtube etc).
One thing I've forgotten was to also get how long it took to discharge with that setting. That was an important data I've overlook, but again, it was a very informal test.
http://imgur.com/a/qfaEI

I'll take some time to read this paper. Might enlighten me a bit more.


-------------------

Hey Unclewebb, I've unlocked my BIOS and I think I found out a way to remove the TDP limit.
The BIOS software there doesn't let me go over the default TDP, but ThrottleStop doesn't complain this time around, when I increase it (previously, it would fall back to the maximum)
http://i.imgur.com/a3dPzU9.png

What's still limiting my processor TDP to a maximum of 47W?

 

Hmm..I see too many 0s with b4..?

 

Hi guys I'm new here.
Intending to get a celeron n3050 dualcore cpu based netbook/cloudbook (4watt TDP passive cooling fanless) to exclusively run a browser tab with flash plugin automated advertisement 24/7
From what I see on benchmarks the n3050 with weak Atom architecture is barely 1/4 the computation power of my current desktop cpu. According to task manager on my desktop PC the automated browser is taking up 1/4 to 1/2 cpu resources.
I estimate the netbook/cloudbook to be barely adequate but I would prefer using TS to force the turbo constantly at maximum to alleviate the bottlenecks when the browser auto refreshes.

The question is ... would ThrottleStop work on the celeron n3050 cpu ? windows 10 ?
There are a dozen budget laptops based on this specific n3050 + win10 configuration coming out so it will be a significant pool of potential users in 2016.
Thanks!~