The Acer ThrottleStop Thread

What to say?

Thank you ?! I think that is not enough!
Any more cool tools around ?! Keep them coming !

 

I only saw this thread today. I have to say to unclewebb that you are well deserving of rep for all your successful hard work done to solve a major Acer (and others) throttling problem. Kudo's to you.

+1 from me. It's the least I can do. Thanks on behalf of all whom you have helped :smile:

(I hate seeing good, giving, informative people stuck with miserable Rep. Burns my tail feathers).

 

Hi there,

Great program so far but for some reason I don't have the SLFM option. SLFM works properly for me in RMclock and CPU genie but not here. Another problem (kinda) is I can only set my VID to a minimum of 1.0375 however i can actually set it down to .925 in the other programs. However the weird thing is when I set it to the min VID in throttlestop, CPU-Z shows my VID at .925 so it seems like a type to me. My processor is the SL9300... thanks for all your work!

 

thendless: Core 2 CPUs use two different formulas for VID; one formula for desktop and a different formula for mobile CPUs. ThrottleStop doesn't realize an SL9300 is a mobile chip so it is blocking access to SLFM and using the wrong VID formula. I'll try to get that fixed up tonight. Thanks for bringing that to my attention. Can you post a screen shot of CPU-Z so I can make sure that I get this right?

Edit: In the CPU-Z specification box it shows what information is written into the CPU. I noticed on some of the SU processors that instead of having SU9300 written in there, Intel only writes U9300. Something like that is probably going on with your SL9300. If ThrottleStop finds L9300 instead of the correct SL9300, then at the moment it won't recognize your CPU correctly. Intel likes playing games like this to give programmers nightmares.

As soon as I see a CPU-Z screen shot to confirm this, it will be simple to fix.

 

Here you go, thanks again!

 

Thee fact that the SL9300 is a low-voltage Core 2 Duo might also be giving the system issues.
Come to think of it, I've never heard of a low- or ultralow-voltage CPU exhibiting throttling issues. They tend to run so cold it's never a problem.

 

ThrottleStop 1.92
http://www.sendspace.com/file/e8ipia

Here's a testing version of ThrottleStop that hopefully solves your problem. Let me know if it works out.

If you can, post another screen shot of ThrottleStop and CPU-Z with one thread loaded with SuperPI or Prime95 or whatever.

 

Here:

One with SLFM on and one with it off. Do those values make sense?

 

The Intel documentation always leaves out a few details so I usually go there first, grab the sSpec number and then head to CPU World to learn some more.

http://processorfinder.intel.com/details.aspx?sSpec=SLB65

http://www.cpu-world.com/sspec/SL/SLB65.html

CPU World says a SL9300 supports Intel Dynamic Acceleration (IDA) which is similar to turbo boost on the newer CPUs. The default multiplier for this CPU is 6.0 and the IDA multiplier is 7.0.

Super Low Frequency Mode (SLFM) is designed to drop the internal bus clock in half from 266 MHz to 133 MHz. This is confusing to software so different programs report this differently. The lowest multiplier possible is 6.0 on a Core 2 chip. When SLFM is active, ThrottleStop and CPU-Z divide the multiplier by 2 and report that.

266 MHz X 6.0 = 1600 MHz

When SLFM mode is active that becomes

133 MHz X 6.0 = 800 MHz

That's hard for software to detect correctly so that will be reported as:

266 MHz X 3.0 = 800 MHz

Your first screen shot shows that you are requesting a 7.0 multiplier and you are also requesting SLFM mode.

Both CPU-Z and ThrottleStop are reporting 3.5 ( 7.0 / 2 ) and that seems to be correct. Your multiplier is 7.0 but SLFM is causing your effective multiplier to be half that much or 3.5.

Your second example is more interesting. ThrottleStop uses timers within your CPU to accurately determine the average multiplier. When IDA mode is active, it is not necessarily active 100% of the time. It can cycle on and off hundreds of times a second so your multiplier will be constantly bouncing back and forth between 6 and 7. ThrottleStop reporting an average of about 6.6 would mean that your computer is using the 7 multiplier about 60% of the time and the 6 multiplier about 40% of the time.

CPU-Z rounds things off and is only sampling your multiplier once per second so it shows 7.0. I think ThrottleStop is giving you a more accurate look at what's really going on inside your CPU.

Another good test to do is to run just a single threaded bench mark like SuperPI. When your CPU is not as busy, it should be able to use IDA mode for a larger percentage of time so ThrottleStop should report a higher average multiplier when that happens.

I don't have any hands on experience with a SL9300 but that seems to be what's happening. Most Core 2 CPUs don't give you any IDA when both cores are active but your SL9300 seems to give you some IDA but not the full +1 multiplier boost. A screen shot when running a single threaded benchmark should be interesting. Take a screen capture about half way through a Super PI run or similar.

You might see a multiplier close to the full 7.0 on one core and closer to the default 6.0 on the second core.

VID looks a little better.

Edit: I just noticed that in your second picture your CPU is not up to full load. I usually test with the Small FFTs option for a consistent load. If you have C1E enabled you can also end up with the multiplier dancing back and forth between 6 and 7 when lightly loaded. That might be part of what you're seeing. You can use RealTemp or i7 Turbo to toggle C1E. I'm going to build that into a future version of ThrottleStop since it's handy when testing.

 

Here are my results with superPI:

 

When lightly loaded, Intel CPUs that support Intel Dynamic Acceleration (IDA) or turbo boost automatically schedule tasks on a single core which allows the other core or cores to enter the C3/C6 sleep state.

CPU-Z refers to the cores as Core 0 and Core 1. In your example, the C0% number shows that Super PI is being mostly scheduled on core 1. This allows core 0 to enter the C3/C6 sleep state. With Core 0 asleep, this allows Core 1 to use the 7.0 IDA multiplier. As soon as Core 0 wakes up to process any background tasks, the multiplier for both cores is limited to 6.0.

The multiplier on Core 0 is averaging 6.0 since it's not getting any Intel Dynamic Acceleration at the moment. Core 1 is averaging 6.8 which shows that it is running at 7.0 about 80% of the time and at 6.0 20% of the time. In this example, the 6.0 multiplier that CPU-Z reports does not accurately show you the true speed of your computer. Many users think IDA is a gimmick that doesn't work because most monitoring software doesn't report it accurately.

80% IDA is a good result. I think you can see that less background activity should result in a higher average multiplier since IDA can be maintained for a longer period of time.

Your second screen shot is the same as before where if you select a 7.0 multiplier and SLFM, you end up with an effective multiplier of 3.5.

I hope this explanation makes sense. If you have any questions just ask.

If the multipliers that ThrottleStop show are jumping around a lot at idle then you need to adjust the Minimum processor state in the Control Panel -> Power Options as well as C1E.

 

Your explanation was perfect unclewebb, thanks a lot for all your hard work and passion, it really shows in your posts. Currently I've replaced RMclock with Throttlestop. It uses much less resources and works just as well if not better. Is there anyway to have the program just startup automatically when windows boots or is that something you plan on implementing in the

 

If you have Administrator rights on your account you should be able to copy a link to ThrottleStop into your Startup folder.

That's the easy method and you can also use the Task Scheduler which works well in Vista or Windows 7. Here's a tutorial I wrote a while ago for RealTemp in case you're not familiar with the Task Scheduler. It's more complex but gives you better control over start up items.

http://www.xtremesystems.org/forums/showpost.php?p=3970161&postcount=3657

If all that multiplier stuff makes sense then the Task Scheduler should be easy to figure out.

Thanks for the feedback so I could get ThrottleStop working correctly on the low power mobile CPUs.

 

I recommend taskscheduler as well.
Microsoft is advocating the use of it and trying to kick the habit of startup folders.

 

Great I will take a look at this when i get the chance. On a completely unrelated matter, I noticed you are from Cochrane in the XtremeSystems forums... I am in Calgary, haha small world!

 

So I was monkeying around with ThrottleStop and IDA, and discovered something interesting.
ThrottleStop was able to detect the extra multiplier (13x) on the busy core, but RMClock didn't!

That's ThrottleStop for you - Canadian made, Canadian grade!
(donation's been sent)

 

weinter: For Windows 7 Microsoft should have made every startup item go through the task scheduler. That's always been a big problem with Windows, way too many different ways for junk to start up at boot up time. Great for virus and spyware writers though.

My favorite program to monitor start up items is Autoruns. These guys knew so much about Windows that Microsoft finally bought them out a couple of years ago.

thendless: Very small world. I grew up in Calgary but ran away to Cochrane 7 years ago. I got sick of sitting in traffic all day. I'm glad they finally built a Memory Express up in the NW. That's a quick drive when I need some more stuff to test with.

TehSuigi: That's what I've been saying. RMClock is a great program but it doesn't properly detect the half multipliers so it won't always properly detect IDA mode either. Thanks for your contribution to my projects.

 

Sent mine as well. I really hope ThrottleStop develops more in the future, there is so much potential for this.

 

Thanks. I need to know what you'd like to see in ThrottleStop. I have a few ideas but I'd like to keep it fairly lean for laptop users. On the things to do list at the moment is optional temperature reporting in the system tray.

I'm going to look into integrating this into the operating system better like RMClock does. Windows has changed this between XP and Windows 7 so I'm going to have to think about this for a while and test a few things first. Hopefully I have the time available.

 

It's not a half-multiplier, it's a full one. I go from 12x (2.4 GHz) to 13x (2.6 GHz), one whole bin up.
One thing I'd request (and one of the reasons I'm sticking with RMClock to undervolt for the time being) is the ability to set VIDs for each FID, and then allow Windows to scale up and down as needed.

 

Read my lips Kevin: P-R-O-F-I-L-E-S.

 

That's exactly what I'd like to add to ThrottleStop. I hope I can get this figured out someday.

And of course I'd like to add some profiles to keep my friend burebista happy. He's been twisting my arm for a long time to come over and help out laptop users. I didn't realize how bad some of them are suffering until after I arrived.

TehSuigi: When you are using RMClock to control your multipliers and ThrottleStop for monitoring, do you get any IDA increase in your multiplier as reported by ThrottleStop or are you stuck at 12.0 when running a single threaded app?

 

ThrottleStop does show the increase in multiplier (IDA does activate) when RMClock is running things.
I'd post screencaps, but both my cores are busy encoding video right now.

 

Ok, ThrottleStop seems to be great tool and solves my Acer 6935g throttling problem too (the rig starts to throttle just after running 10 minutes of GTA4).

But can you make an installable package out of it that could set it automatically start on windows startup? Would be much more usefull when one wouldn't need to click ThrottleStop to start every time before game is started

Thanks for the great soft though!

 

Hi,

As unclewebb posted previously, add it as a scheduled task. Works like a charm

HTH,

Dave.

 

I don't understand why one couldn't adjust the minimum processor state setting in the power options:

http://www.dslreports.com/forum/r23639106-WIN7-CPU-throttling

I have a i7 720QM. For the plugged in state, setting to 50% keeps the multiplier at 7x, as reported by cpuz and throttlestop. At 60%, it sits at 8x. 70% = 9x, 80% = 10x, 90% = 11x, 95% = 12.x, and 100% will peg one of the cores into turbo mode, whereupon cpuz reports a multiplier of 16x, 18x, or
20x. Why then would one need Throttlestop (with all due respect to the author of said program)

 

KookyMonster: You need to read some more about what this issue is all about. TehSuigi's post will explain the problem better.

http://forum.notebookreview.com/showthread.php?t=403988

Acer is using a feature called Chipset clock modulation to throttle and reduce the performance of their laptops. Unfortunately adjusting the Minimum processor state won't do anything to fix this issue.

 

Me happy
Lack of profiles are the only thing that keeps me with RMClock.

 

Hi guys, not been on here for a whille but what a nice suprise to logg on and see. I'm at work now so i can't test it for myslef yet (allthough do %100 belive that it will work like a charm). But I do just have one quck question though, is it recomended to use rmclock's undervolting with throtlestop to at least try to cut down on the cpu temps a bit or can throtlestop undervolt also?

Anways thank you in advance to unclewebb for what i'm sure is an amazing program, and also a big thank you to tehsuigi and the other guys on here for there continuing efforts to help sort this problem out

Edit: not read all of this thread incase it has allready been mentioned, sorry if it has!

 

The advantage of RMClock is that it is better integrated with the operating system compared to ThrottleStop. The disadvantage is that it doesn't fully support the newer 45nm CPUs with their 0.5 half multipliers.

ThrottleStop can be used to adjust VID and FID but it's not yet as flexible as RMClock is. Depending on what CPU you have and its default multiplier, you might want to stick with RMClock and then only use ThrottleStop to control Chipset Clock Modulation. ThrottleStop is a fairly small and lean program so it won't consume too many system resources. With Chipset Clock Modulation taken care of, you don't have to worry as much about core temperatures but under volting is still a good idea.

I look forward to your test results DaRush. For VID and FID you can run RMClock or ThrottleStop but don't try to get them to both control FID and VID at the same time or else they'll get into a fight.

 

Nice 1, cheers for the heads up unclewebb. Can't wait to get home and stress my 6920G to it's limit hehe *fingers crossed*

 

I cant get this to work I click on the download it gives me a file and it keeps giving my errors once I click on the link what exactly do i have to do thanks!

 

ThrottleStop 1.92
http://www.sendspace.com/file/excyi8

I just uploaded it to sendspace.

When you download it you might need to unzip it first so it can find the driver.

Can you be more specific about the error you are getting?

Hopefully TehSuigi's post can help you out.

http://forum.notebookreview.com/showthread.php?t=403988

 

i keep getting the error

windows cant access the specific device,path or file you may not have the appropriate permission to access the item.

help ? lol

 

You have to unzip and move the file folder to a location that you have access to and you have to run this program with Administrator privileges. Putting the folder on your desktop is a good place to check it out.

It's a powerful program that hacks (I mean adjusts) your CPU at the Ring0 level so you need to be in an account with Admin rights or know an Admin password so you can run it with Admin rights.

You're the first person having trouble. We might have to deduct your Rep Power if you don't get this working.

 

Yesterday I gave throttlestop a try. I downloaded the 1.92 version.
Than I started prototype (this game was unplayable on my 8930g). To my surprise I was able to run it at normal settings with 1920 * 1080. But after a minute of 5 the machine froze! Tried rebooting and thought I try the intel burn test. After 5 minutes my machine just shuts off instantly...

Than I thought I'd try orthos and the same happened.

For some info.
8930g with 1.14 with slic 2.1 bios.
x9100 cpu
windows 7 home premium 64 bit
4 gb ddr3
9700m gt 512 mb
4gb turbo memory.

In throttlestop i check clock modulation 100%, chipset clock mod 100%, set multiplier 11,5, log file and the minimize on close box.
Don't really know what is going wrong here.

I've also checked the log to see if it was throttling but al the values (cmod% and chip% were 100% all the times. Can someone enlighten me?

 

And your CPU temps in full-load are?

 

Stupid me!!! Didn't even check that... I'm at work right now so I cannot check it here as I leave my lappie at home... will do so this evening and report back here... you could be right you know... perhaps temps are exceeding the safety limits so it shuts down.

 

At a first glance that's the cause. Now you don't throttle anymore but your core temps should rise in games/IBT/Orthos.
Kevin's RealTemp should tell you the truth.

 

Thanks, I will give this a go this evening.

 

Confirmed 100% working on my 8930G!
Thanks a lot unclewebb! After months and months of struggling to game properly with this laptop... you have provided the working solution!

 

@spaanplaat: Yeah, that's the other, more legitimate reason why your CPU might throttle; it's actually reaching those dangerous temperatures.
Might be worth a look into cleaning out your fan intake and exhaust.

 

Has anyone tried measuring power consumption with a Kill-a-Watt yet on these laptops when stress testing or gaming with ThrottleStop enabled?

Depending on configuration, some users might be getting close to the 90 watt limit of the adapter which could also cause shut down problems.

spaanplaat: For your X9100 just enter

TJMax=105

in the ThrottleStop.ini configuration file and TS will report and log your approximate core temperature. The sensors aren't that accurate but they are better than nothing.

 

I still cant get it to work I download it then extract the files and ran it as administrator and still get that error. Im not doing something right lol.

 

Halo360Fan: Do you have a ThrottleStop folder on your desktop? Does it have ThrottleStop.exe in there along with 4 WinRing0 files and a ThrottleStop.ini configuration file?

Are you using any exotic anti virus / anti spyware type program? It might be something like that which is blocking the WinRing0 driver or maybe ThrottleStop itself has a bad reputation since it can make your computer run hotter by allowing it to run at full speed. It seems to work fine for everyone else.

Are you in a limited account? Even if you are, if you run it as an Administrator it should work.

 

it was kaspersky it had it as restricted i got it working going to test it now

 

Thanks for getting that figured out. I'll add a note about Kaspersky to the documentation. I'm not a big fan of antivirus programs that prevent you from using your computer the way it was designed to be used.

 

Wow, just wow.

I sold my 8930G about a year ago because of the lag-problems in games. I remember I posted so many postes on this forum, but noone could fix it at that time, and now a genius makes this program. I'm speachless, good job!

Luckily I don't have throttle problems anymore with my new Acer 6930

 

Ok, riddle solved...or not... it seems with throttlestop I cannot undervolt as much as with rmclock. With rmclock I could go as low as 1,125 v but with throttlestop it must be 1,215 v. It passes intel burn test at max stress level and runned orthos for 12 hours straight without errors.

Seems it is not getting to hot... but even for the reccord I will post a little from the log from realtemp.... the temp isn't changing at all!!
That can't be right... right??

DATE TIME MHz C0 C1 GPU LOAD%
02/01/10 23:04:30 3059.06 33 32 44 100.0
02/01/10 23:04:35 3059.06 33 32 44 100.0
02/01/10 23:04:40 3059.06 33 32 44 100.0
02/01/10 23:04:45 3059.06 33 32 44 100.0
02/01/10 23:04:50 3059.06 33 32 44 100.0
02/01/10 23:04:55 3059.06 33 32 44 100.0

02/02/10 03:46:25 3059.06 33 32 41 100.0
02/02/10 03:46:30 3059.06 33 32 40 100.0
02/02/10 03:46:35 3059.06 33 32 40 100.0
02/02/10 03:46:40 3059.06 33 32 41 100.0

02/02/10 08:04:30 3059.06 33 32 40 100.0
02/02/10 08:04:35 3059.06 33 32 40 100.0
02/02/10 08:04:40 3059.06 33 32 41 100.0
02/02/10 08:04:45 3059.06 33 32 41 100.0
02/02/10 08:04:50 3059.06 33 32 40 100.0
02/02/10 08:04:55 3059.06 33 32 40 100.0

02/02/10 18:04:30 3059.06 33 32 42 100.0
02/02/10 18:04:35 3059.06 33 32 42 100.0
02/02/10 18:04:40 3059.06 33 32 42 100.0
02/02/10 18:04:45 3059.06 33 32 42 100.0

But fortunatelly I have not experienced throttling since... I think... or perhaps not...

Just one question for you guys, how come all the values in different programs show different values when they should all be the same?

Watch the red boxes... I thought it should always be 11.5x multi and 3.059Ghz or something very close but in fact I see the cpu throttles anyway? Despite the fact I checked the box in throttlestop to always be 11.5x. Also sometimes cpu-z shows a 3x multiplier. See what I mean?





So what's happening here....

Ohh and b.t.w. I did not test gaming after tests.... will do so now I'm done

 

You might have one of the broken versions of RealTemp. The X and QX processors are different animals and I was just working on fixing RealTemp up to better support those two when I got side tracked by ThrottleStop.

The Control Panel -> Power Options -> Minimum processor state setting can control your MHz at idle. So can enabling C1E. Windows 7 may not be respecting those settings though on some motherboards or laptops.

Can you run Prime95 Small FFTs on both cores for just a few seconds until the load comes up to its maximum. Can you post a screen shot of ThrottleStop and CPU-Z while that is running?

The internal timers that RealTemp and ThrottleStop use are extremely accurate on Core i and regular Core 2 processors but they are not approved for use with QX processors when you manually change the multiplier in the bios if you have that option. I don't think that is happening in your case but with some more info I'll let you know what's going on.

A 3.0 multiplier is CPU-Z's way of saying that SLFM, super low frequency mode is being detected. ThrottleStop is not correctly identifying your CPU as a mobile CPU so that's why the SLFM option is hidden and why it is using the wrong formula for VID. I'll get that fixed up for you.

I wish I had a bag full of CPUs so I could get them all right but I don't and the Intel public documentation is somewhat limited.

The Link Width that is being reported as x1 by RealTemp is troubling. Try running GPU-Z and see what it says. If that is true that can kill gaming performance. Hopefully a few more Acer users can test for this and post if GPU-Z shows anything funny. Typical link width is x16. CPU-Z should also be able to report this. I don't believe that is a RealTemp bug since I am getting that information directly from the Nvidia driver.

RealTemp has an option that you can add to the INI file.

MSRMulti=1

This will use an older method to determine the multiplier that doesn't rely on the internal timers.