The ThrottleStop Guide

ok so it does not mean my cpu don't goes in lower states just that throttlestop is not reading it

 

I wonder if the ES versions of the sandy bridge CPUs have them hardcoded?

 

I think all of the Sandy Bridge mobile CPUs have the maximum TDP locked except for the 2920XM. ES or OEM shouldn't make any difference.

For the 2920XM, it's up to the laptop manufacturers to write a bios and decide whether to leave these unlocked or to set the lock bit so they can't be adjusted after you boot up. For the Alienware M18x, Dell charges an extra $75 bucks so the unlocked 2920XM really is unlocked.

 

hello eveyone.. in kinda newbie in this kind of stuff..
so, to make thing simple i have m11x r2 i3 330um.. so, i use this throttledstop, what is the maximum FID?

 

The Core i3-330UM does not support Turbo Boost.
The default maximum speed is 1.20 GHz ( 133 MHz x 9.0 )

http://ark.intel.com/Product.aspx?id=49021

ThrottleStop can accurately monitor your CPU but it can't be used to increase the amount of Turbo Boost because your CPU doesn't use that.

 

means that its useless for i3 330um to use throttlestop?

 

I have a Dell Latitude D620 with an Intel Core Duo T2500. It likes to throttle at about 80 degrees, which is super annoying, so I was thrilled to find this utility.

It doesn't seem to work for me though. I notice that it's a chipset throttle, which TStop nicely reports to me, but setting the Chipset Clock Mod to 100% doesn't seem to help. Here is a screenshot:



I should note, however, that I can successfully change the chip% initially. But once the CPU heats up and the chipset initiates its throttling algorithm (gradually stepping down the chip%), I can't make the chip% any higher than the chipset wants to make it (although I can still make it lower).

I am running Windows 7 Pro with the power option set to high performance. I've tried ticking and unticking the box, no help. Any ideas? Is the Core Duo not supported?

Thanks for your help.

 

In theory, ThrottleStop should support your T2500 but there are some laptops where a hardware request for chipset clock modulation will take precedence over what software like ThrottleStop wants so ThrottleStop won't be able to help you out.

When you first start ThrottleStop and Chip is being reported as 100.0, are you able to use ThrottleStop to adjust that from 12.5 to 100.0 without any problems? That would prove that ThrottleStop is able to control Chipset Clock Modulation but the problem is likely what I first mentioned.

Can you send me a log file when this first happens? Check the Log File option and the More Data option and have ThrottleStop set up like you do in the above picture. The log file might be big so you can upload it to MediaFire or SendSpace and then post a link here.

You shouldn't need an account at either of those file hosting sites.

Have you tried using ThrottleStop to lower your core voltage yet? This will significantly reduce the amount of heat these 65nm processors put out and might help prevent this problem from happening. When voltage testing I usually run a program like Prime95 Small FFTs and load both cores and then while the program is running, I reduce the voltage one step at a time until the program reports an error. You might want to wait a couple of minutes at each step if possible. This will give you an idea of how low you can run the voltage reliably. You can test for a longer period of time after you come up with a reduced voltage setting that you think will be adequate.

If reducing the core voltage doesn't solve this then the next thing to try is removing the heatsink and redoing the thermal paste. Some laptop heatsinks don't have enough retention strength between the heatsink and CPU so you have to get creative and find a way to snug it up. Doing the little things like this can help temperatures quite a bit.

 

Thanks so much for the quick reply.

Yes, that's right. Before the throttling algorithm kicks in, I can adjust everything just fine using ThrottleStop. Also, at the beginning of the throttling process (before it's all the way down to 12.5%), I can adjust chip% to be lower, but not higher, than the throttled value.

A log file of this behavior is attached here, with the settings identical to the above picture. Once it gets up to about 78 degrees, chip% starts to gradually step down, all the way to 12.5% (the log file cuts off at 25%).

I have tried this, and it does help a little.

Thanks so much for your time and your help.

 

Just curious if anybody figured out a way to allow Asus notebooks to benefit from Throttlestop, aside from voltage regulation. Been a while since I followed this thread and I hate to read through the whole thing.

 

gubanichplague: I thought that maybe I could adjust the ThrottleStop anti-throttling algorithm a little but your log file makes it clear that ThrottleStop is not able to stop the chipset clock modulation throttling on your CPU.

How low have you been able to reduce the core voltage? Your CPU might be able to run reliably at ~1.15 volts which should make a difference. I have a 45nm CPU in my D830 and it runs great at 1.05V without ever getting anywhere near the throttling point. Reducing voltage and redoing the thermal paste could be enough to avoid this problem. If I can think of anything else, I'll send you a test version of ThrottleStop in a few days when I have the chance but it probably won't help.

smokinokie: Asus makes quite a few laptops. If you don't mention what laptop model or CPU model or what problem you are trying to fix then its difficult to answer your question.

 

ThrottleStop can only unlock the multiplier on the Core 2 Extreme CPUs like the X9000, X9100 and Quads like the QX9200 and QX9300. Intel made sure that none of their regular non-Extreme CPUs could be multiplier unlocked and that will never change.

 

Yeah, it looks like that's the case. So it seems like with this CPU, there is no possible way to circumvent the hard-wired throttling algorithm -- would you agree with that assessment? Is it possible that upgrading (or downgrading) the BIOS would make a difference? (I don't know if the BIOS has anything to do with this, though)

After an extensive test, it looks like I can run as low as 1.0875V without any errors (after ~35 minutes of testing -- obviously I should test for much longer to make sure). And it runs about 10-11 degrees cooler at that voltage, so that is pretty great.

Is there any danger at all to undervolting like this? From what I've read, it seems like there really isn't, given that it can be shown to be stable for a long time, which apparently it can. So why the hell do they make these CPUs to run at much higher voltages than they would ever need? Couldn't they advertise a way better battery life if they didn't?

Also, I have to admit I don't completely understand the relationship between the core voltage and the multiplier. I've noticed that when I am on the balanced power setting and the multiplier switches on an as-needed basis, the multiplier and core voltage basically switch in tandem, in between the extrema of (1.2625V, 12.0x) and (0.9500V, 6.0x). EDIT: Okay, I think I understand why this is the case -- basically you need more dynamic range when you want to do calculations fast because it takes time for the circuits to equilibrate.

One question related to this: is there a way for me to keep the "balanced" power setting such that the multiplier and voltage switch dynamically, only with a maximum allowed core voltage? EDIT AGAIN: I think I figured out how to do this with RM Clock.

Thanks again for all your help.

 

It's possible that a different bios version will use a less aggressive throttling algorithm. I like using the latest version.

CPU under volting has been around for a long time. As long as you don't get carried away and your CPU is stable, you won't have any problems. My T8100 has a default IDA voltage of 1.20v but I've been running it at 1.05v for quite a while. Power consumption is proportional to voltage squared so that's almost a 25% decrease when the CPU is loaded.

CPUs can degrade over time so Intel uses a little more voltage than necessary to be on the safe side. The average consumer doesn't have the time or know how to test and fine tune their core voltage so I'm sure Intel does this to reduce any less long term problems. I can only imagine how much energy world wide this one decision has wasted.

ThrottleStop only works correctly when using the Windows high performance profile. On the newer 45nm CPUs, I didn't see any advantage by using the intermediate multipliers so the Power Saver feature in ThrottleStop only uses the lowest multiplier and the highest multiplier. The sensitivity of this feature can be fine tuned with the PowerSaver C0% setting in the Options window. I found doing this allowed the CPU to spend more time in the lower C3/C6 sleep states so power consumption using only two multipliers is pretty much the same. Here's an example of that.

http://forum.notebookreview.com/6369641-post5060.html

I haven't done any significant testing on the earlier CPUs like you have so if you need more control, RM Clock is a good choice. Reducing your core temperature by 10C might be enough to eliminate the chipset clock modulation problems you were having.

 

Just FYI, the undervolting stuff we're talking about works fine for me on the balanced setting. Can't speak for the throttling settings obviously.

Is it the "minimum processor state = 100%" or the "system cooling policy = active" setting that's important?

Yeah, after playing around with it for a while, I think that's pretty much true for mine too.

Random question: is there any particular reason why you chose a default value of 35% for that? Is there any reason I shouldn't set it to like 95% or 100%, therefore guaranteeing that it's always in the lowest possible state?

It sure is. After trying to figure out (for years, not kidding) exactly how to fix this problem, it's nice to finally have it figured out -- and to know exactly how my computer is working. I really appreciate your help with it.

 

When using ThrottleStop, I use the Windows high performance profile and set the minimum processor state to 100%. If you don't do this, ThrottleStop and Windows can fight against each other, trying to control the CPU. If you want or need a low idle multi, I would still leave this set to 100% and then I'd use the ThrottleStop Power Saver feature. On my T8100 I just use the highest IDA multiplier and 1.05v. It runs great and doesn't burn through my lap.

The 35% number was mostly a number off the top of my head. Use whatever number you like. I don't use this feature because I prefer a high multiplier at idle. Do you see any drop in your core temperature at idle when comparing a high multi to a low multi? That's usually a good way to prove if a setting is saving you some power.

I was a little worried when I saw your log file. Glad to help you find a Plan B that is working for you.

If you ever pull your laptop apart, try reinstalling the CPU heasink with some IC Diamond 7 thermal paste to improve those temps a few more degrees.

ICD7

 

May I ask why you prefer a high multiplier at idle?

I'm finding that (6.0x, 0.95V) gives a core temp that is is 2-3 degrees lower than (12.0x, 1.10V). So I think I'll be using ThrottleStop with PowerSaver C0% set to 90.

I will definitely do this the next time I tear it apart to replace the hinges or something.

 

I prefer the high multiplier at idle because in my 45nm Core 2 CPU, I have found zero difference at idle in the core temperature when doing this. I know this is different than how the original 65nm CPUs operate like you have. Have a close look at the two graphs in this post.

http://forum.notebookreview.com/6369641-post5060.html

Whether the FID and VID are high or low doesn't seem to make any significant difference to power consumption when the CPU is idle. This CPU spends over 95% of its idle time in the C3 or a deeper sleep state where it uses a much lower VID internally so I think that's why there isn't a big difference.

You can use the Windows Performance Monitor if you are using Vista or Windows 7 to measure battery power consumption or to determine what sleep state your CPU uses when idle.

The ThrottleStop PowerSaver option was designed for your 65nm CPU but it doesn't seem to make any significant difference for the 45nm Core 2 mobile and Core i mobile CPUs that were released after yours.

 

Hi was having an issue.

We have a Windows 7 Standard User that will be using a laptop with Throttlestop.

We created the task under the admin account and we can see that Throttlestop is running but it is not showing in the Windows 7 Standard User's Systray.

Is there some sort of additional setting when setup in this manner that will allow Throttlestop to be displayed in the systray?

 

Hi, i have an issue with my Intel X9100 please see this 2 pages thread http://forum.notebookreview.com/har...-most-performance-x9100-laptop-processor.html.

What is the best configuration in throttle stop for my processor?

Thanks for this great program unclewebb.

Regards

 

For a small boost test I also want to force the C3 state to my i7 2600k cores.
With my current bios the cpu never goes into C3; with all consequences for
the boost...

What program was used to force the cores into the states?

 

BDI-IT: ThrottleStop is not multi-user friendly and I haven't played with this for a while in a Limited account. You should only use it in one account at a time and if you want to run it in a limited account or an account that uses UAC, try setting it up using the Task Scheduler exactly as outline in this example:

Task Scheduler

ranziger: Dufus likes to write his own software but he generally keeps it for personal use. The three programs in your picture are his creations.

Are you sure? What ThrottleStop reports for C3% time is for the entire CPU package. It's still possible that individual threads are going into C3. What are you using to monitor for this? Can you post a screen shot of ThrottleStop and explain what you're trying to do? If the bios doesn't set up C3/C6, I don't know if you can force this with software after you boot up. Does Turbo Boost not work at all on your i7-2600K?

 

No matter what I do, in Perfmon and in ThrottleStop no other value
than 0.0 for the %C3 is shown. Because this is the necessary state
to allow the cpu to distribute the power to the other cores I only can
apply the lowest turbo with all cores active.

Without ThrottleStop I have no turbo. So thanks for TS, so I can
set the multiplier to 35 instead of 34. When I completly disable all
except of one core in the bios I got the 38 multiplier.

So I think it is more or less a bad implementation of the bios.
Also the Tmonitor and the Intel TB Monitor 2.0 are saying
that my cpu is not supported...

 

Run CPU-Z and click on the Mainboard tab and see what chipset your board is using. Multiplier overclocking is not supported on motherboards using the H67 chipset. You need a P67 chipset to overclock.

Does ThrottleStop show anything for C6%?

There is a button that shows C1/C3/C6/C7. What is that set to and is it locked?

I think on any board you should get the 38, 37, 36 or 35 turbo multiplier when 1, 2, 3 or 4 cores are active. If you run a single threaded benchmark like 1 thread of Prime95 and use the Task Manager Set Affinity function to lock it to a single thread of your CPU, does ThrottleStop show an average multiplier over 37 on that thread?

Edit: If you want to find out if your first two cores are using C3 or C6 then you can use my MSR Tool and monitor these two registers.

C3 - 0x3FC
C6 - 0x3FD

MSR Tool
http://www.mediafire.com/?myjkxzkzzmd

With your CPU idle at the desktop, enter either 0x3FC or 0x3FD in the MSR Number box and click on the Read MSR button about once every second. If the value in this register is not changing then that Core is not spending any time in C3 or C6.

 

hello,sir.
can you tell me How to set multiplier by programming?

 

If you write data to MSR 0x199 you can change the multiplier. What you need to write depends on what CPU you have. Have fun playing.

 

No, also 0.0%. And yes; with the chipset I made a mistake. I'm one
of those, who don't read the spec's carefully and bought a K-processor
together with a H67 chipset. But I don't think, that this causes the
problems with the boost.

When this is the C1-button between the BCLK and the DTS button,
then it is locked.

No, always 35. And after a lot of reading and searching I believe this is
because of the 0.0%C3 / C6

I also used your MSR-tool and the rw 1.4.9.7 for monitoring the mentioned
msr's. No change with or without load; no effects, when setting the affinty
to a specific core.

 

The C1 button is usually at C7 on most Sandy Bridge based boards. When that button is set to C1, that prevents the CPU package from using the C3 or C6 sleep state. That's a bios bug.

If you checked those two MSRs and neither one is changing at idle then individual cores are not being allowed to use C3 or C6 either which will prevent turbo boost from working correctly.

With the H67 chipset, you won't be able to adjust the multiplier in your 2600K but it should allow turbo boost to work correctly. It's up to the motherboard manufacturer to fix the C3/C6 sleep state bugs in the bios so turbo boost can be fully used. It's possible that they did this on purpose to prevent you from using turbo boost but if they did, that was a very poor design decision on their part. Blocking C3/C6 reduces performance and increases power consumption.

 

It is good to hear that from another side... I'm in contact with my manufacturer. I hope they will fix it. The only purpose to prevent using the boost I can imagine is, that they are concerned about the peak in the power consumption. I have to say that I'm using a Shuttle-System, so the PSU is very small. The power in general is enough. But I have figured out scenarios were peaks in the power consumption of the GPU (6970 @ 500W PSU ) causes the system to crash. Out of the box the system comes with a 300W PSU. So I can believe, when they know that the PSU can only "slowly" increase the necessary power, that they suppress the C3/C6-states to prevent crashes. I think it wouldn'd be the first time, that deeper C-states
are causing crashes...

edit: just got an answer from the support (within 1 day!). They are already informed about the c3-State problem
and are currently working on a new bios...
/edit

For now I think there's nothing I really can do. Only if there would be a tool
for forcing the C-states to specific cores... like the one dufus uses.
But it only would help me, if I can do this "on the fly" in windows. If there
is a way to do this it would be nice when you let me know.

 

thank you sir.
but...can you give me the demo asm code?
or how to use WinRing0 to do it...
thanks again...

 

my cpu is core 2 x9100.
if i want to set the multiplier to six.
how clould i write the code by asm?
or use WinRing0.
thanks.

 

I wanted to thank the developer for this program.
The fact that the multiplier is now locked on the max level on my t7700 finally gives me near-instant reaction from the computer, thus making the use of the computer far more enjoyable.
Without it I keep experiencing lag, probably because of the slow switching between the cpu power states.

I have a request, though: every time the computer comes back from standby EIST is enabled and the cpu starts changing c-states again. Would it be possible to have ThrottleStop enforce the chosen options after the computer comes back from standby, thus keeping that option off at all times?

 

elevul: Are you using the maximum Intel Dynamic Acceleration (IDA) multiplier on both cores? I call this Dual IDA or DIDA. If you use Profile 1 for this then go into the Options window and put a check mark beside Profile 1 and when you resume from Stand By, this should use the maximum multiplier again and it will disable EIST. If that doesn't work, give me a few more details.

ranziger: I'm pretty sure that your C State problem can only be taken care of at the bios level. At least they know about this issue and are working on it. This might be an interesting tool to play with.

http://biosbits.org/

Sorry colorful, you'll have to do your own homework. WinRing0 is no longer available from the author but if you do some Google searching you should be able to find it somewhere. It includes full documentation of each function and explains how to use it.

Try using MSR Tool to read and write different values to MSR 0x199.

MSR Tool
http://www.mediafire.com/?myjkxzkzzmd



Here's a hint. The 22 in MSR register 0x199 is the VID code and the B in that register is the multiplier.
0xB hexadecimal is 11 in decimal so this is how to request the 11 multiplier. If you want the 6 multiplier then change the B to a 6. If you play around with different values in ThrottleStop and then read this register, you will also be able to figure out a formula for converting the VID data into voltage.

Volume 3B of the Intel documentation has a big list of various registers and what they contain in Appendix B. They also leave out information about many of the important registers. That's when you have to do some trial and error testing. Have fun.

http://www.intel.com/Assets/PDF/manual/253669.pdf

 

It worked. Thank you.

 

thank you, sir.
i will do it myself.
by the way, can you tell me your E-mail address?

 

i cannot find the WinRing0 documents, i have found it for 3 days.
can you give me a url?
thank you, sir.

 

3 days? I went to Google and typed in WinRing0 download and I was able to download the complete WinRing0 package including documentation from the third link.

Winring0 download - free download - (4 files)

 

i am sorry, sir.
In China, we cannot use google because some Political reasons.
If i used, i maybe be killed today. We only can use BAIDU.
And i cannot open your link in my Network environment(i am in an university).

 

so, can you upload the [WinRing0 with documents] as a Accessories?
thanks.

 

Hi.
I have a T9300. In order to undervolt as much as possible, do I need to test and set voltages for every multiplier, or I just need to set voltage for the highest multiplier that I use? If former, in case of T9300, should I test by 12.5,12,11.5,11....or by 12.5,11.5,10.5....? Thanks

 

Is it a good idea to set "Turbo" multiplier when i am gaming and game like that for few hours? My cpu seems to be getting to 86+C (which is hot)

 

iu1nguoi: Set the multiplier as high as it can go and run 2 threads of Prime95 or whatever your favorite stress testing program is and then with the same settings run a single thread of the same test program. Think of ThrottleStop as a simplified version of RM Clock. If you want to test every multiplier and set a unique VID for each multiplier then you will need to use RM Clock to do that. I found on the newer 45nm Core 2 CPUs that there is not a lot of additional power savings during normal use by doing this so ThrottleStop concentrates on taking care of the maximum VID voltage and that's it. Check out the two pictures in this post and you will start to understand my reasoning for doing this.

http://forum.notebookreview.com/6369641-post5060.html

Lieto: Intel CPUs are rated to run 100% reliably up to the thermal throttling point which is typically 100C. Intel sets the thermal shutdown temperature to 125C to 130C for most of their CPUs.

86C is hot and I wouldn't want to put my thumb on that CPU core but it is a perfectly safe core temperature for an Intel CPU. If it wasn't safe, Intel would have lowered the thermal throttling point years ago but they didn't. If it hits 100C, the CPU will throttle and slow down. ThrottleStop can't prevent that and that should be enough to keep your core temperature from getting anywhere near the thermal shutdown temperature so no worries.

I'm assuming that you have a Mac Book Pro. The real problem seems to be an inadequate cooling solution. I was helping one user the other day and his CPU was getting up to 100C during normal use and then throttling and slowing down. This limitation prevented him from fully loading his laptop. Kind of like paying for a V8 engine for your car but you only get to run 4 cylinders at a time.

If you want to do some testing then try running the wPrime benchmark. If it hangs when it first starts, hold the Shift key down to bypass the hardware detection stage. A properly designed quad core laptop should have no problem running a wPrime benchmark at full speed while all 8 threads are fully loaded. If your CPU gets up to 100C during this test and starts to thermal throttle then that's a sign that either the heatsink is not installed correctly or the manufacturer cut some corners at the design stage.

colorful: WinRing0 is powerful stuff. If it is blocked by your country or university then there is nothing I can do about that. I don't have the legal right to redistribute WinRing0. You can try contacting the author if you can find him.

OpenLibSys.org

 

thank you....after checking 2 cores, do I need to do so for every single core?

i just realized that if I put all threads of Prime95 to 1 single core, that core will run at overclock speed (13.5x, I did Dual IDA as your way), but if I put all threads to all cores (2cores), it will only run at hightest normal speed (12.5x). is it normal?

 

Any ideas how can i boost performance without shutting down ths system?
Atm it turns off after 15 minutes of gaming Everything i did for now is setting turbo on.

 

Any Idea if I can use it on AMD Processor?

I wanna overclock the AMD V105(1.2Ghz) because it slow but the graphic card is great. the processor becomes a bottleneck to the performance.





.

 

Either your CPU heatsink is poorly installed or it is poorly designed. Pulling your laptop apart and redoing the thermal paste with some ICD7 will help but it probably voids your warranty.

Some laptops with these new Quad Core Sandy Bridge processors are under designed and can not run at full speed without overheating and throttling. With ThrottleStop you can Disable Turbo and perhaps lower the multiplier so the CPU puts out less heat. That might be enough so your laptop is at least usable for gaming even if it can't run at full speed.

iu1nguoi: Dual IDA is when a T9300 locks both cores at the 13.50 multiplier. You need to have EIST disabled to do this. If both cores are active and your multiplier drops down to its default of 12.5 then you are definitely not in Dual IDA. Can you toggle EIST off and on? Many laptops have the EIST bit locked so ThrottleStop will show EIST grayed out.

Fully loading and testing each individual core with Prime95 is a good idea.

DeadPool88: Sorry, ThrottleStop is for Intel Core CPUs only.

There is an AMD program called K10Stat but I have no idea if it's good or bad or if it will help you out.

 

after disabling EIST in BIOS, I can toogle it.
so as you said, can't I undervolt 13.5 multiplier?

 

To under volt you have to do that first and then disable EIST. Toggle EIST on, set your voltage and then right click on ThrottleStop and you should be able to send the CPU into Dual IDA mode. Profile 1 is designed to be used for Dual IDA. If you use any other profile, you will have to be manually in charge of this.

SpeedStep (EIST) needs to be enabled for FID or VID changes.

 

2 unclewebb
You seem really experienced with this kind of stuff.
Can you give any tips on what multiplier should i try or maybe should i try to undervolt it?

I have — 2.2 GHz quad-core (2720QM) Intel Core i7 Sandy Bridge with 6 MB on-chip L3 cache

Its new so i doubt changing thermal paste will help, i ll try it as well as using external cooler ZM2000 which i have at work. However its still 20C in my room but in a month it will prolly jump to 30C so i need to think ahead, apparently turbo mode wont be safe or do you actually think changing thermal paste (getting some better paste) can make dramatic change? (i can do it in apple care center or something).

[edit] btw can i control my fan speed in throttlestop? Maybe thats part of the reason because they are either rather silent in general or not spinning at max speed.