The ThrottleStop Guide

hi there.....great app!!

how can this be used to stop ''power limit throttle'' ..???, sometimes i cant seem to get rid of the throttle...

im guessing the adjustments i need are on this page


or this one.......



sorry im a bit of a throttlestop nub....

just wanna get rid of the ''power limit throttle''..???

 

Long term, I would like to have individual power limits for each profile. Projects like this seem simple but they can become very time consuming because I do not have access to a computer with unlocked power limits for testing purposes. I will see what I can do this year.

@GrofLuigi - It is possible that the wrong CStateLimit value was in your INI file from a previous version of ThrottleStop or perhaps you updated your bios and the default value for this changed. I agree that this part of ThrottleStop is a little confusing at the moment. I need to change it so it clearly reports the current value your CPU is set to as well as the value that you would like to set this to and show whether the register is locked or not.

@skunk1 - First of all, what power limit throttling? Turn on the Log File option in ThrottleStop, go play a game or crunch some numbers and show me what your CPU is doing. I would turn off Intel XTU while using ThrottleStop. Copy and Paste some data to www.pastebin.com so I can have a look at the problem. For settings, uncheck the Clamp option. There are multiple power limits. ThrottleStop shows that one of these limits is adjustable but there might be another power limit that limits your CPU to its TDP value of 45 Watts. I hope not.

I just remembered that for Skylake based CPUs, I need to double check to make sure that I turned off Limit Reasons monitoring that is going to the log file. The Limit Reasons program and data does not support Skylake CPUs.

 

thaks kindly unclewebb......

i think ive managed to quash the throttle...... HWinfo is now able to detect my fans, ive since set them up .... chips cool, no throttle..

i think what was happening was the fans never fired up until 60c and even then id say it would only hit 30% ... so the heat & power would spike it would hit 70-80c instantly ''power limit throttle'' kicks in.... game over!!

ive since stressed it and all seems well...

 

There are about 401 sites in this thread and I couldn't find anything that would really help me, so I think I'll ask here.
I own the infamous Iconia W700 with the Intel Core i5 3337u (I got it for 250 euros) and would like to reduce throttling as much as possible.
Thing is: I have no idea how to use throttlestop at all, especially for the W700. I'd like to undervolt it as much as possible and also make it throttle less (I'm not sure how, as I noticed that the CPU throttles at 60°C).
I'd be happy if anyone could help or post a link to some helpful information, as everything I found was two years old.
Also, did anyone release a modified BIOS for the W700 that changes the target temperature?

Thank you very much in advance!

 

so far ive managed to get a stable 35x multiplier when all 4 cores are active without thermal or power throttling on 4710hq. But recently i noticed some XTU benchmarks where people get 36x or even 37x, how did they do it? I cant set 37x for 3 or 4 cores in XTU, and setting them in throttlestop ( xtu uninstalled ) doesnt help as well, they just stay on 35x max


Here is example with tweaked voltages



Here is another example with stock voltages

 

@ Uftherr - The Core i5-3337U is a low power CPU. Your CPU is throttling at only 60°C due to power consumption, not heat. Long term, a 3337U cannot run beyond 17 Watts. I do not think that there was ever a solution for this. You might be able to do some minor tweaks and improvements by using ThrottleStop but as far as I can remember, you cannot totally fix this type of throttling. You cannot use ThrottleStop to undervolt a 3rd Gen Core i CPU.

@vidumec - The maximum multiplier for a 4710HQ when 4 cores are active is 35. ThrottleStop lets you select the 37 multiplier whether 1, 2, 3 or 4 cores are active but to obtain the 37 multiplier when 4 cores are active, you will likely need a modified bios. Going beyond Intel's limits is possible with some MQ CPUs but it might not be possible at all with your HQ. Are you sure that people were getting the 37 multiplier while the XTU Benchmark was running? Sometimes people will snap a screenshot just after a benchmark is finished. At that point, it might be showing the 37 multiplier because only one core was active. Post some links so I can have a look.

 

@unclewebb
Thanks for the answer. I somehow was expecting that..
What are those minor tweaks? To me, the tablet is pretty useless in it's current state.
And I also read something about being able to balance CPU and GPU. Right now, the tablet runs the CPU at about 1000 MHz and the HD 4000 at 350MHz. Is there a way to change it to lets say 850-500?

 

@unclewebb


from http://hwbot.org/benchmark/xtu/rankings?hardwareTypeId=processor_3620&cores=4#start=0#interval=20

two even got beyond 3700 MHz
Maybe its somehow inaccurate or im missing something? The highest score i could get was around 700 as well

 

Hey unclewebb..

My i7 3621qm is making a high pitch when in power saving states.

I've read around and saw something about CStateDemotion but I can't quite figure out how to set it accordingly? I've seen "CStateDemotion=3" but does that go in the .ini file or what?

Should be noted that I can remove the whine completely by using a regedit tweak that allows me to disable the "processor idle" state in Windows power settings. However this naturally wrecks batterylife and increases cpu fan and so on, by using the absolute max clock...

Isnt there some middleground that could be found?

EDIT: Ok, Read some more.. it's the C7 interupt response limit I'm looking for, yes? I tried setting it to 1 and it sounds like the pitch is gone... or reduced.. how big an impact does not using these c states have on battery?

 

What does EDP mean

 

@vidumec - Here are the details about aquastevase's 4710HQ.

http://i.imgur.com/g8MreRI.png

It looks like HWBOT reports the 1 core multiplier even though the CPU has 4 cores active during the XTU benchmark. The benchmark itself also shows the highest multiplier which is misleading.

If you want to learn about a trick that lets you go WAY BEYOND Intel's limits, contact 0.0.

http://hwbot.org/submission/2839754_0.0_cpu_frequency_core_i7_4700mq_4752.7_mhz

Tell 'em unclewebb sent you. Is this possible on a 4710HQ? Maybe.
0.0 discovered that by using an early microcode, the multiplier is wide open, similar to an Extreme CPU without the Extreme cost. Not sure why more people were not interested in this trick when he shared it on NBR.

@tb227 - EDP stands for Electrical Design Point. I think EDP means current limit throttling is holding you back.

http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/page-295

@Kocane - The regedit tweak usually gets rid of all of the C States. This creates extra heat and usually disables full turbo boost so I would avoid doing that on a laptop. On my 4700MQ, adjusting C7 IRL can be used to disable the Package C States. The individual core C States still work so there might be a slight increase in power consumption and heat but nothing major. A laptop that whines can get very irritating so I would keep using that trick. Thanks for reporting a new use for C7 IRL.

 

I am trying TS on the surface pro 4 and am noticing some things.

Firstly, I'm no expert on what the new voltage settings do. I've only noticed that dragging all bars left still decreases temperature and allows longer turbo boost, but also don't understand what the new integrated voltage regulator box is for. This concerns me only because on another system, moving the voltage bars below 50 crashed but that system is not with me and TS has been updated with new voltage controls in the meantime.

Is this basically crash protection for large overvolts?

 

Yes, that's right. Current limit be what EDP lighting up would signify.

3700MHz is the maximum single-core OC limit for a 4710HQ.

3.3GHz 4-core turbo, 3.5GHz 1-core turbo.
3.5GHz 4-core turbo max OC, 3.7GHz 1-core turbo max OC.

Above 3700MHz would be bclk tuning to get a bit above. Likely it was just a bit above, and not say... 3800MHz.

 

Toshiba R30-A，
The original data：
Current Limit：29.4 A

4700mq must set the Current Limit 256 to keep the frequency does not reduce；
Is it safe to set 256？

 

Thanks

 

What CPU model do you have? All CPUs are unique. One CPU family might be happy with a - 50 mV offset while another CPU family can run reliably with a -100 mV or -125 mV offset. Have you tried adjusting VCCIN? I am not yet sure what CPU models use VCCIN so feedback about these new features is always appreciated. I do not yet have access to any Skylake documentation or hardware for testing purposes so I have been programming in the dark lately.

That sounds like a good thing. Decreasing CPU voltage will decrease power consumption and this will decrease heat output. Intel Turbo Boost is limited by power consumption so undervolting provides another benefit. Decreasing power consumption will allow your CPU to use more Turbo Boost so it will run faster. Undervolting should also let you run longer on battery power. Definitely a win win situation. The lower you can go the better as long as you do not lose stability. If -125 mV is stable then you can switch the Range button to the 250 mV setting and try to reduce the voltage further.

Just curious. Have you tried using Intel XTU? Is your CPU supported? How does it compare to ThrottleStop for voltage features?

I assume you are talking about the new VR Faults and VR Efficiency Mode settings. I do not fully understand these myself. I posted an explanation earlier.

http://forum.notebookreview.com/threads/the-throttlestop-guide.531329/page-398#post-10168611

These settings might not work on your CPU model and even if VR Faults does work, it is probably not necessary on a U CPU. Who knows though. Maybe one of these settings could be very useful on a U CPU. I just program this stuff. It is up to users to share with others in the forums about what works and what doesn't. The default for both of these seems to be checked and Enabled.

It sounds like disabling VR Faults might allow a person to overclock more before power consumption becomes a limiting factor. I think unlocked desktop K series CPUs or unlocked mobile CPUs would be the most likely CPU models to benefit from this setting. Intel XTU includes these so I decided to include them in ThrottleStop too. The advantage of using ThrottleStop to adjust these is at least ThrottleStop is consistent and accurately displays whether they are Enabled or not. This is especially important after using Stand By since XTU seems to randomly forget some settings after a Stand By / Resume cycle.

 

This is the Skylake Core i7-6650U. It's the "best" among the 15W (or if this tablet is indeed TDP-down configured when unplugged, 9.5W) new Skylakes. I have also tested a Core-m7 (also Skylake, 4.5W fanless) and reported previously, but that one tops out at about 65mV undervolt (for any of the levers except GPU, but the moment I try to do something with the GPU, it dies despite passing the XTU benchmark. Odd).

I will do this. What bits of info are most interesting with this feature? I could report temps during benchmarks at various settings--and/or their effects on other settings (like how far I can undervolt, performance of benchmarks)?

Ah, hadn't seen that switch. I'll try lower and get back to you. Trying -250mV did indeed crash the tablet just now -- so the setting is clearly working at least!

I have, yes. It has the same features, but with certain caveats (it seems TS is more permissive). Firstly, it doesn't let me decouple core voltage from cache voltage. Whenever I drag or set one, the other is updated to the same. When I tested in TS, I didn't see significant temperature decrease if I just adjusted one of the voltages either -- but maybe I didn't do enough nuanced testing.

Another caveat is adaptive voltage -- XTU doesn't let me set my own custom adaptive voltage at all. The moment I enter a voltage of my own, the adaptive voltage button becomes grayed out and static is selected. In TS, When I try to combine Adaptive with my own entered voltage, it uses default voltages no matter what. Voltage offset is the only slider that works with adaptive voltage (in both XTU and TS).

Third caveat -- the Turbo boost settings do not seem to work in TS, and changing them in TS is not reflected in XTU (tested by uninstalling XTU, changing the settings, reinstalling XTU, and checking). The other settings TS touches are reflected in XTU properly (except when not supported -- like when I try to set core and cache voltages differently, XTU shows one of the two -- I think the higher -- for both). However (here's the strange part), when I use XTU to set turbo properties (like disable short turbo, changing the watts from 20 to 50, etc), the change is shown in TS and actually works (the SP4 will benchmark for longer in turbo and then start to throttle by heat only, according to XTU's graph of throttle reasons).

Thanks, I'll try to disable or tweak them and check temps and OC-ing stability/performance.

Interesting you mention standby. I have had trouble after installing XTU to enter standby -- the device would reboot after entering standby after a few moments. Using TS, this has not been an issue. Curious -- maybe XTU is doing something bonkers (like forcibly setting/unsetting some things) when standby is triggered.

 

Hey everyone. I'm wondering if I could get a little help here.

Behold, my dinosaur, an m1710 (flashy e1705 inspiron) with a 7950 "go" gtx. Yes, it is old.
I have a core 2 duo t7600g, and throttlestop allows overclocking no issue, however games actually STILL run quite decently on it surprisingly, but I'm having extreme CHIPSET specific throttle issues. Like throttling down to 1ghz cpu. Pretty hard throttling. This is with fans set to Max "4200+ rpm".
It will always occur when there are typically a lot of particles or special fx being rendered. I have attempted every configuration in throttlestop I can think of, from disabling speedstep, to bd prochot, to selecting only chipset or cpu clock modulation.
Nothing really works. The only relief the throttling has had has been downclocking the gpu, but hence it's too slow then.
The weird things are, when adjusting the cpu clock, it has to be below stock to guarantee no throttling (Bullsh**)...and if it is a gpu overheat scenario which it might be, why wouldn't bd prochot work in that case?
I'm at whit's end trying to figure it out, hence the forum post.
Any help par chance?

And yes....I will get more cooling installed, I monitor my own temperatures and don't require the cpu to throttle.
It's a real shame, at 2.66 ghz cpu, and 675/800 gpu it's pretty much the perfect sweet spot for any game that can still run on it....if it didn't throttle to a crawl of 1ghz at times when anything epic happened for a moment.
Downclocking the gpu or cpu isn't acceptable as a fix, and yes I will be installing more cooling, however, this is more of a "why won't the program seem to work as advertised".
Core 2 Duo T7600G
M1710 A07 i945PM chipset
Temps if it matters...with clean fans
gpu @ max load 65-75C
Cpu @ Max load 50-60C
I can try to get the MSR data up here at some point too if that will assist.

 

First: Have you repasted CPU+GPU with good quality TIM?

And cleaned out the fans, ducts and fins?

 

AS5 and copper shims. Couldn't be cleaner yes.

 

hey unclewebb, i came across this post " http://forum.notebookreview.com/thr...ier-tdp-tdc-values.514239/page-2#post-6694990" that throttlestop has the ability to auto go between profiles, i looked into it a bit but cant seem to figure it out.

how would one go about doing that?

 

@matchbox2022 - When you are having a problem, post a screenshot of how you have ThrottleStop setup and include a Log File when using those settings so I can see what the CPU is doing. Include GPU temperature monitoring in the log file. Are you using the original power adapter and how many watts is it rated for? If you have upgraded your power adapter, make sure the bios fully supports it. Usually if a Dell bios does not recognize an adapter, you will not be able to get the full rated power out of it. Your laptop might be starving for some more juice.

Go into the Options window, check off the Alarm feature and then you can ask ThrottleStop to auto switch profiles based on CPU or GPU temperature. This works but the problem is that if the two profiles that you are switching between have a large difference in performance, the change in core temperature will be drastic and your laptop will sit there and constantly switch profiles back and forth every second. I never use the Alarm feature. Intel CPUs do a great job of managing themselves at the hardware level where CPU performance can be rapidly adjusted hundreds of times per second to keep the core temperature from getting out of hand. The result when Intel thermal throttling kicks in should be a much smoother decrease in performance.

The desktop Skylake CPUs do not use VCCIN anymore so I was just curious if your U CPU uses this. If you adjust this to 1.80 V and hit Apply and the monitoring table on the right hand side does not change then I should disable this feature since it doesn't work.

The XTU programmers have access to a lot more documentation and devices to play with compared to what I have access to. Core and Cache voltage might be tied together at the hardware level in your CPU. When I am not 100% sure about a setting, I try to program TS so it is as flexible as possible. It wouldn't be a surprise if some voltage adjustment does not work correctly in TS.

It can be difficult trying to compare XTU and TS because they are both accessing the same registers in the same CPU. There are situations where you can make a change in TS and this change is shown in a register within the CPU but XTU ignores the change and pretends that it didn't happen. Even restarting XTU doesn't reflect the values that are in the various CPU registers.

The safest thing is to use XTU or TS but not both at the same time. There are still some occasions where you might need XTU to unlock something that ThrottleStop does not have access to. During Stand By / Resume, I think TS has the advantage. TS applies its settings after the Resume is complete so this might give the CPU a chance to settle in before voltages are adjusted.

It is great to hear that TS gives you some control over the voltages in the 6th Gen U processors. Thanks for the feedback.

 

Thank you so much for your assistance and the prompt response, here's what I've got. Also, the MSR CPU Auto Throttling and BIOS thermal controls disabled according to the thermal / tech analysis mode on this dell. Can't seem to grasp why I can't just tell it STOP with your program, since the nvidia card will simply halt itself if there's an overheat. I have speedfan running, as you can see the temperatures are not that high, mid 60s. fans on 4500rpm. 156.55xp 32bit forceware drivers for the gpu. I think just stock 2006 era dell chipset drivers for that. It's an m1710, almost identical to the e1705 inspiron 2006-07 era.

Have also included a cpuz Registers dump. Lots of info in there and a lot of hexadecimals I don't grasp but it might have something to do with what your program can control / whats going on.
The PS is 130 Watts oem dell, newer flatter version, the bios does read it as 130 watts, and it's what the system calls for...although even if it were underpowered (I highly doubt, else I'll get 200 watt dell oem one) can't throttling still be disabled? (risking psu damage obviously).
Your program does not read my gpu temperature at all. Though I can overclock, volt and throttle manually at will.
My temperatures do increase dramatically to 75C for the gpu if I lower my gpu clock, and it will take longer to throttle the cpu in response to high temperatures. However, throttling still happens and Im trying to make it not occur. No idea but I'm assuming somehow there is something telling the cpu the gpu is increasing it's temperature at an alarming rate (even if it hasn't climbed there yet) and throttling. However I am trying to disable this troublesome feature, especially for some of the more cpu intensive games vs the gpu ones. (thought it was bd prochot, which I've played with on and off a ton with no results...unless your program physically can't use that with my laptop for whatever reason) since I can do things for cooling and clocks myself.

 

I've been rolling with it set to "1" for a while, since it feels like it's the best result. But there's still some whine. Nothing to do?

 

@matchbox2022 - Unfortunately, there are some throttling schemes that take precedence over what ThrottleStop wants so ThrottleStop might not be able to fix this problem.

When your laptop is idle and not throttling and you switch Chipset Clock Modulation on the left side from 100.0% to 87.5% to 75.0%; does the Chip column on the right hand side also change from 100.0% to 87.5% to 75.0%? I am just trying to confirm that ThrottleStop is writing the correct info to the correct register.

Can you try upgrading to ThrottleStop 8.00? It might handle the Chipset Clock Modulation stuff a little differently. There is also an INI option that might help out. It adjusts how often ThrottleStop checks the value and tries to change the value in the Chipset register.

TimerInterval=4

If this key is not found in the ThrottleStop.INI file, it defaults to 25. At the moment, the minimum value this can be set to is 4. Add the above line to the INI file and see if it makes any difference. A smaller value tells ThrottleStop to check this register more often. If there is any improvement at all, I will adjust this so you can try using lower values like 3, 2 or 1. At the moment, any value less than 4 will be changed to 4.

I feel your frustration. Your temperatures are great so if this throttling was not holding you back, you could increase the VID and probably get another 400 MHz out of your CPU. If you cannot solve this throttling issue, you will probably need to overclock a lot less, both the CPU and GPU, just to avoid triggering throttling.

ThrottleStop gets temperature data from the GPU driver. If you were using a newer driver, ThrottleStop might show your GPU temperatures. Have you tried using any newer driver versions. There might be something in the driver itself that is contributing to this problem.

@Kocane - In the ThrottleStop C States window, is the Package C State Limit button grayed out and locked? If it is not locked you could try adjusting this to C1. If that is locked then there is not much else you can do. I think some manufacturers cheap out on capacitors and you can end up with irritating whine issues like you are having.

 

Thanks so much for your help.
So I raised the graphics on crysis, and it's giving my old ass gpu a run for it's tits and you can clearly see the cpu is throttling to keep this old thing from blowing up basically, max fans it's still crawling toward the 80C mark, that's when I lower gpu clocks so it won't lock up. It will throttle it seems when it's close to a gpu lockup (though I've had it lockup before with no throttling whatsoever).
Can't seem to disable it though I know better. I have tried your ver 8 as well as inputting the change in the ini file. And can't say there's been a huge change, the throttling is still very reproducible.
I might try the drivers since I hate these ones, but the older ones I used before gave me the same throttling behave as these do. Can't fathom newer drivers throttling less if that is indeed the cause. Given the fact throttlestop can't see my gpu temp does that in essence mean that it can't see whatever command is going from the gpu to throttle via the chipset?

So I'll try them eventually and update when I have some more time with newer drivers and clean install.
I CAN control the chipset throttling manually, I just can't override whatever behavior is making it occur on my own, not without downclocking either the cpu or gpu (I leave one overclocked, the other better be underclocked else there will be throttling in almost any game, I use crysis since it's super demanding and almost without fail produces throttling even at stock cpu frequency).

 

My biggest issue with updating the driver is Im unsure which ones will allow gpu clock changes on the fly and which won't aka locked via any software The ones I have currently are the last Im aware of that allow it, anything later and yes I get some more data but that feature is unavailable and thus I'm stuck with if I'm approaching high gpu temps to just let it kill itself. Underclocking on the fly is far more important for the safety of the system.

 

Yeah it is indeed greyed out. What a shame. It's not as bad as my old HP back in 2007 though (disabling C1 back then I think) helped completely but it seems like with this PC I'm just gonna have to endure it.

 

@Kocane
It's too bad RMClock went out of development, but this is an idea for a feature to add to throttlestop since I think it's become the official CPU controller thesedays. If there's any programs out there still like it, I know for a fact I had this issue almost to a Tee with my cpu in low power states, I think it had something to do with a voltage regulator, it eventually went away playing with some of the advanced options in CPU settings with RMClock.
One I didn't really like since it "pretends" to max out the cpu with empty cycles, but did in fact work was "Run HLT Command when OS is idle" it completely eliminated the high pitch noise while I looked into other solutions.
So maybe something similar, a way to keep your cpu at lowest multiplier when idle and just fill it with empty cycles. Here's a check you can do,with the speakers muted, run something a bit taxing on the cpu but not enough to make speedstep kick in like music and see if the whine goes away.
I have NO clue if RMclock will even work with these cpus it's so dated, but it might be worth getting if only to get rid of the whine while at idle.

 

d

Thanks for the input! RMClock is indeed dated.. The HLT option was greyed out but it seems like a good idea.

I also tried what the person wrote about in one of the last posts here:
http://forum.tabletpcreview.com/thr...for-good-on-core-duo-please-read.3516/page-11

Didn't seem to work though... or.. maybe.. I'll give it a while to try. I don't know how setting the idle promote threshold to 100% is gonna affect performance/battery life though.

EDIT: seems like it stopped the whine lol

 

Where exactly did you input the idle promote threshold, is this with xtu or a windows setting somewhere?
Just want to know in case someone else has an issue.

 

Who knows, could possibly be done with Software as well as removing power throttling for Haswell / Broadwell Q/H/M/U/Y although the power one's a bit risky if running on battery.

Unfortunately as soon as a question or two is asked that might take a minute or two of peoples time or the person needs to do some work on their side then no one is interested, well except for maybe one or two.

 

If there's a way to do away with edp limit on my 4900MQ; I'm game . Don't mind a modded bios; strip all its microcode updates and flash? Have an ES anyway, so dread the day Intel will strike with a new, 'improved' update and cripple every ES/QS on the planet ... Not sure how to prevent Intel/MS from patching it right after boot, though ... is merely no-LAN, no-wifi and no-auto-driver-updates sufficient?

Also upgraded the battery pack and the TS benches (1024, too) run just as fast on battery as they do on adapter, only difference is that limit reasons claims 'Power' as well as EDP. There's no reason to; they can draw 8A easily, so with 6 cells there's 175W to throw around (at 25% capacity left), yet the adapter is only rated for 120W (max. socket draw is even less). Of course, the battery's pcb has a protection circuit (prevents overdraw), but that doesn't appear to kick in with the stock 47W. Besides, there's one or two thermal sensors in the battery pack anyway.

For something completely different; couldn't get TS to run at startup consistently, so used a delayed start along with a vbs to make it open in tray only (no pop-up window) and enable 'on startup' using a blank password. Not sure why it doesn't play nice as a regular startup, though ... but in case anyone else suffers the same annoyance:

Spoiler: ThrottleStop - run on startup

Spoiler: content for the registry

Code:

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa]
"LimitBlankPasswordUse"=dword:00000000
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Policies\System]
"EnableLinkedConnections"=dword:00000001

Spoiler: content for the vbs

Code:

Set WshShell = CreateObject("WScript.Shell")
WshShell.Run chr(34) & "C:\Programs\Hardware\ThrottleStop\TSstart.cmd" & Chr(34), 0
Set WshShell = Nothing

Spoiler: content for the cmd

Code:

timeout /t 10
start /B C:\Programs\Hardware\ThrottleStop\ThrottleStop.exe

 

While ThrottleStop starts up just fine every time on both of my servers, I'm sure that will be useful for others here. Thanks!

 

@t456 EDP can be current or PL3. Try making your current limit 256A or more.

Afterward can you do a Linpack test showing power draw. Sorry, no modded BIOS but may be possible with just software with caveats.

 

With the regedit tweaks listed in that thread, linked in the post I sent. Altho it felt like I haddent heard any whine for some days, it's suddenly back, seemingly. The issue is driving me nuts.

 

Both are locked in TS. Perhaps XTU? Will need a restart, but it's running an x-day process atm (edp limited ...).

So how can we replicate this enthralling story then? The max. TDC is killing performance and assumed the TDC 'unlimited' was the result of the pre-production microcode. Don't mind flashing a modded bios; already tested several, bricked one and fixed it again with soldering iron. No big deal; eeprom is easily accessible.

Spoiler: Core i7-4900MQ ES

In the spirit of sharing information; was also considering a dual bios mod (not this design, though) and place the switch somewhere convenient (a relay-only would not work). That way you could test ad infinitum; flash mod -> brick -> throw switch -> reboot to dos -> throw switch -> flash bricked bios. Since the bios is mirrored in memory by then, it is fine to turn off the good bios and activate the bricked version.

 

Current limit is just a BIOS setting. HWiNFO will show 256A or more as unlimited and perhaps that is what it really means as Martin is privy to a lot more info than us mere users. Less than 256A will show what it is ie a setting of 255A will show 255A.

 

Is there nothing to do, aside from disabling idle (massive fan noise and temps etc), that can prevent CPU whine? Something that'll bring the PC out of this power-saving state. Something to run when plugged in, if nothing else, if it hits batterylife hard.

 

I vaguely remember for C2D whine it was C4 that needed to be disabled, not all C-States. If there's not an option in the BIOS perhaps powercfg has an option as C-States are generally driven from the OS.

 

I can only imagine that Unclewebb knows and would've made it possible to deal with if possible. Sadly, my BIOS offers jack**** in regards to this matter.

 

There's a couple points to make.
On my phone so short form.
If registry worked once then os is saving a idle state profile which is often overridden as a new current control set. Therefore whine comes back on reboot when Smth else says that's not what I have as default and rewrites to registry.Likely cchip set.
I haven't looked into it. Registry setting ultimately will likely be the answer.
Need to encourage CPU to not be idle. Run throttle stop disable c states. Run music in background muted. Unmute if whine still there ;p. Run an infinite loop program on one core. Use Intel xtu to find functions. If whine ultimately vrms try adjusting idle clocks to higher volts.
Finally. Just get older / get some healthy hearing damage. If mine whines still I don't know it does .

 

Rmclock had a enter c3 on c4 command mode. In regards to it supporting c2d. Useless for CPUs past 2009.

 

Look up any free single affinity programs which act like system idle to a given CPU is use inputted by user.. Wish I could code one! Disable c States likely best shot in meantime. If u under volt then heat won't be too bad.

 

Have you looked for such a tool?

I've kinda noticed that the high pitch sound isn't there/not as much when not on AC. Even tho power settings etc, are the same on both "plugged in" and "battery". Does that make any sense? Shouldn't it enter these power saving states just as much when on battery?

Disabling idle/c-states just generates too much heat and noise, it's sadly a no-go.

 

LoadTester is a small tool I wrote for testing purposes. It lets you fully load a single thread in increments of 10% from 0% to 100%. You can use the Task Manager - Set Affinity option to lock LoadTester to a single core if you need to but you probably won't have to do that. It is kind of the opposite of Prime95. It can fully load a thread without creating a huge amount of heat. It might help you keep your sanity.

LoadTester can be found in the RealTemp 3.70 download at TechPowerUp.

http://www.techpowerup.com/downloads/2089/real-temp-3-70/

 

^^^That's probably your ticket!

 

Definitely gonna try this when I get home in a more quiet environment. Like I said, it doesn't seem to happen on battery (or I could endure it when on battery) so having something that might not be too kind of power-saving, while on AC, to combat this issue, would be fine. Thanks

EDIT: Seems like I need to load the cpu quite a lot before the whine actually stops.

 

Just out of curiousity....put your ear up to your AC adapter...is it coming from that?

I think it's a voltage regulator problem then with voltage not being as constant as from battery. Obviously try playing with your idle voltages to see what happens. And though you're loading up a core at least it goes away and frees the idle. It actually puts load on the system though as opposed to a "fake" load which you're right kinda sucks.

 

Yeah I thought about that initially but it's definitely coming from the PC. My laptop is very quiet (doesn't help lol) so I can easily put my ear up to the fan or there around and hear the noises.

I think I'd rather have some "load", as long as it doesn't slow things down or heat things up too much, than having to listen to this whine. I'm really sensitive to highpitched noises