The ThrottleStop Guide

Tons of info to digest, many thanks.

 

Hello...i am inlove with throttle stop....with it i could reduce the performance for my i7 2630 beast....but i wanted to ask is it safe when i use clock modulation and chipset clock mod on 75 percent...to reduce performance.So is it safe for the cpu?It is working it is working fine and with reduced heat but giving me enough power for games.Thank and give opinions -Is it safe>?

 

Hi all, new here. With the help of guides found on this forum (ThrottleStop and The Undervolting Guide), I was able to recently get the most out of my Acer Aspire 6920G.

All seemed pretty straightforward, for ThrottleStop it was just simply changing performance settings (minimum processor state to 100%) and ticking Chipset Clock Mod. Noticed FID multipliers for both cores locked at the highest (10.0), don't really know how I did this as I remember when I haven't turned it on it was jumping around, and I didn't even tick Set Multiplier. More on this later.

RMClock was simple as well, and came after setting up ThrottleStop 4.00. I have undervolted from 1.2500 to 1.0500. Temps are down. Hurray.

Now, I thought I could reproduce this feat on my wife's HP Pavilion dv4 with a Core 2 Duo P8600 (2.4 GHz) running Windows 7 x64 (just like my Aspire 6920G with T5800 Core 2 Duo). I was wrong.

My issues (I have changed my performance settings already):

1. I tried to lock the Multiplier to the highest by ticking Set Multiplier then 9.0, but even after enabling, the FID still "jumps". How do I lock it to the highest multiplier? Side question, the selector on Set Multiplier allows me to go up to 9.5 but once I do this the laptop always crashes. Why?

2. On my Aspire 6920G it is always running at the maximum, EVEN with Set Multiplier unticked. How the heck did I do that?

3. What kind of throttling does my wife's laptop have? For the Aspire it was chipset based, but what should I tick for this laptop to disable throttling?

4. I am happy with the way my Aspire 6920G is running now with the FID locked to the highest but I want to know how I did this. I tried turning off TS but even in monitoring mode the FID was still locked. Is it supposed to behave like this? I thought the multiplier should "jump" once you turn off TS?

5. Last question, is it ok to have your CPU running at the highest multiplier or does this have any drawbacks? I am returning temps of 52 C at high load after undervolting. Is this ok or should I just unlock my FID multiplier?


Sorry for the noobish sounding questions but knowing the answers to my questions would help me sleep better at night.

Thanks all.

(I didn't see this thread was still alive before making this post and made a separate thread on this...)

 

BlazerMFT: I put an answer for you in the other thread you created.

http://forum.notebookreview.com/8080414-post2.html

Clock modulation works quite well in the new CPUs and is a great way to control their maximum speed and power consumption and heat. Don't bother with Chipset Clock Modulation. Only use the regular Clock Modulation.

 

(I'm still alive...)
I've been thinking here... let me explain my case.

I have a T3200 cpu (pentium dual). It runs at 2ghz (12x166) at a default of 1.25v. I am capable of undervolting it to 0.9875v at 12x. The lowest voltage possible for that cpu is 0.900v.
With a power meter, I found that using the 0.9v I could save an extra 1.5W (not much, but it's something). But the only way to do that is to use the lowest multiplier (6x -> 1ghz) with "power saver" option.
Well, I know that this cpu can run at 10x multiplier with 0.900v and having it running at 6x won't give me any more power savings than running it at a higher speed but with the same voltage. It will only increase latency and possibly a minimal performance hit.
So what I was thinking is that throttlestop, instead of idling to the lowest multiplier possible, it would idle at the multiplier and voltage that I choose (new feature ...). Since it would be running at higher speed, cpu won't need to change the multiplier when lightly loaded and run at 0.900v more time than jumping straight to 0.9875v because of a little load at 6x
Than it would jump between 10x and 12x multiplier (0.9 and 0.9875v)

What do you think?

 

I think that's a good idea TigTex but I don't know if I will ever get around to doing it. I think RM Clock works OK on these older CPUs so why not give it a try?

Maybe I can come up with a quick INI option for ThrottleStop to handle this situation. Too much football on TV tomorrow so I probably won't accomplish too much and then after that it is back to work for a couple of weeks. Project ThrottleStop is mostly suspended at the moment but I will see what I can do.

 

Well, I haven't tried rmclock because it is not as lightweight as throttlestop. It impacts performance a little, and I like your work

It would also be nice to have a check box to hide the throttlestop icon when any option is enabled in "notification area" just to save some space on tray area because clicking on any cpu/gpu/mhz icon it will open throttlestop as well.

Keep up the awesome work

 

I'm looking into BITS to see if it can be used to allow some overclocking on SB CPUs. From the sound of it, if it doesn't work, nothing will.

UPDATE:

Even in BITS MSR 0x610 was still locked. I tried changing MSR 0x1AD for higher multipliers, with the idea of them loading a new CPU microcode to try to reset the CPU to the higher turbo frequencies, but BITS said my ES CPU was unrecognized, so I could not load a microcode. Maybe overclocking would be possible through BITS on a retail 2920 or 2960?

I emailed the author asking for help.

 

Hi unclewebb.. I am running ThrottleStop 2.89 since a while now, and I have it to start along with windows. TS allows me to overclock my Extreme CPU, over 3GHz because I can increase the voltage. I am really happy with TS. So far only two issues that I don't know how to solve,

1) Sometimes at windows start, TS display a message "Could not open winring0.dll".. this issue is really annoying because TS will not start untill I restart my machine.

2) Other issue is sometimes at windows start TS displays a message "overflow.. something else"... this issue is not that bad because TS starts anyway...

Is there any way to solve the issues? specially the "could not open winring '??

 

ThrottleStop needs the WinRing0 driver to get legal access to the CPU's registers. Unfortunately I didn't write WinRing0 so there is nothing I can do about it not opening up correctly sometimes. Small programmers like myself can't afford to get a Ring0 driver signed for x64 compatibility.

You can try using either version of ThrottleStop from my signature. They have more features and are significantly more efficient than TS 2.89.

I can't remember with what version but sometime during development I increased the amount of time ThrottleStop sits and waits for the driver to be found. That might help. Windows is big and bloated and not nearly as multi-threaded as it should be so sometimes it gets bogged down at start up, especially when using a slow hard drive. The Task Scheduler is the best way to add ThrottleStop to your start up sequence.

That info should be in the second post of this thread.

http://forum.notebookreview.com/6865107-post2.html

 

Started using throttlestop today to oc my xpsm1730 x9000 cpu, must say this program really does it goood! Got till 3.8ghz at 1.3875v. I was wondering the VID is that limited by the software or is that what is limited by the motherboard itself?. Asking because i can go to 4ghz and run at 1.5v which is the max but if i load up then it would bsod but a few more volts i think i can run it full 4ghz.( I ran rmclock and saw that i can select up to 1.6xx v but rm doesnt read higher than 19x)
thanks again for this great piece of software!

 

Max VID you can select is determined by the CPU. Seems to be 1.5V for X9000/X9100 CPUs.

Worth noting that a X9100 is being cpu pinmodded from 266->200Mhz and then used in a 965PM system. Advantage there is the E0-stepping X9100 runs substantially cooler than a X9000, they're more plentiful and cheaper on ebay, they are a better candidate for use in newer PM45/GM45 systems. See RickiBerlin's writeup at http://forum.notebookreview.com/hp-...n-owners-lounge-part-2-a-457.html#post8125412 if interested.

RickiBerlin as also got a quad-core QX9300 booting up in his PM965-equipped HDX9000. Just need to figure out how to run it with all 4 cores.

 

Intel uses two different VID tables. One for mobile CPUs and a different table / formula for their desktop Core 2 CPUs. If RM Clock is showing a hair over 1.60 volts, that's only because it is using the formula for desktop CPUs on your mobile CPU. There is a way to switch that so it reads your CPU correctly in RM Clock.

ThrottleStop reads these CPUs correctly and lets you feed your CPU the maximum possible voltage. 3.8 GHz is the practical wall for most of these CPUs. Beyond that you need full voltage which usually creates too much heat so you end up with the BSOD.

Great to hear that you finally discovered ThrottleStop. It helps breathe new life into older laptops.

 

Do you know if it's possible to run any Montevina CPU in Santa Rosa based notebook? I would be interested in running medium voltage CPU such as P8700.

Also, I've heard that newer Socket P found in quad-core notebooks is different from that found in older ones. Isn't that true?

I think I found a bug in ThrottleStop. I have a T9300 rev C0 CPU and ThrottleStop reports wrong DTS value. It looks like Your program "thinks" that Tjmax temperature for my CPU is 100*C, which in fact is 105*C (according to info found in the internet). In result, ThrottleStop reports temperature lower by a 5*C, than any other tool.

Other than that, I found ThrottleStop very handy. At the beginning I've been confused, but Your documentation clarified everything. I can't wait for a newer laptop with i5/i7 CPU to try Turbo settings.

One another question: does anyone happen to know if Penryn CPU is physically lower than Merom? Previously I had T7700 in my 6910p, and when I swapped CPU I found that temperatures went high. It turned out, that cooling wasn't firmly connected with CPU. When I lifted up the CPU by mounting it 1mm higher in socket (not an easy thing), temperatures dropped by 20-30*C in heavy stress.

For all of you overclocking your CPUs in laptops: try running FPU Stress test in AIDA64 System Stability Test. I've been using many programs heating CPU, but this one gives highest temperatures (on any Core 2 Solo/Duo/Quad CPU. I don't know about newer Core i3/i5/i7 CPUs).

 

I have Throttlestop starting up with Task Scheduler. It does start as I can see it in the Task Manager, but it's not showing in my System Tray.

Please help, thanks.

 

forteller: When you swapped CPUs did you delete the previous TJMax valule in the ThrottleStop.ini file? ThrottleStop should be using the correct 105C TJMax value for your T9300 but if it finds a different value in the INI file, it uses that instead. If you exit ThrottleStop, delete that item from the INI file and then restart ThrottleStop, it should use the correct value. The 100C value was probably in the INI file from when you were using your T7700.

hulawafu77: Are you using the Windows show all icons feature? When Windows hides system tray icons, it seems to cause problems for ThrottleStop. I don't have a fix for that problem.

Did you follow the Task Scheduler method exactly?

http://forum.notebookreview.com/6865107-post2.html

The other problem when booting up is that it is possible for ThrottleStop to start running before the task bar exists. The program starts like you found but it can't be accessed. For that problem you might want to try delaying the start of ThrottleStop. I added an INI file option to try and help with this issue but make sure you created the task properly first.

DelayedStart=10

That would create a 10 second delay before ThrottleStop starts.

 

big Thanks to UncleWebb again for the wonderful program

I'm still using ver 3.0, runs solid. So was there anything significantly changed so that I should update the version ?? I see we're up to ver 4.00 now ?

forteller: Intel's own stability test program is called LinPack, it heats my CPU more than Orthos, try it out.

also, the fact that I'm thinking in PLL overclock from 200 to 266MHz is all nando's fault Just let me figure out the cooling ...

 

unclewebb, I want to thank you again for creating the ThrottleStop software which without it I couldn't enjoy playing games on my G53 laptop.

As I live in a middle-eastern country where Paypal and other online paying services are blocked I can't donate a beer or else I would have donated a cold beer to you.

So all I can do is saying thank you very much and I wish you luck and success in your life. You are the best.

 

I didn't know that ThrottleStop stores this value in ini file. Everything is fine now. Thank you very much

 

miro_gt: Most of the changes between version 3.00 and version 4.00 were new features for the Core i CPUs but I think version 4.00 is more efficient than version 3.00 for all CPUs. The new Hotkey Manager might be useful if you want to quickly switch profiles from the keyboard. You should be able to do this while gaming or benching.

Farhang_Maiden: There is more to life than money. Knowing that there are thousands of happy ThrottleStop users all over the world is payment enough for me.

forteller: Before Intel started writing TJMax to a register in each of their Core i CPUs, users could never agree what TJMax was supposed to be set to. Most monitoring programs allow you to adjust this with an INI or similar option. Glad to hear everything is OK.

 

Love your software and have a few questions. I'm using version 4

I have an Asus G73-SW which has a really annoying throttling issue.

1. An asus forum suggested TS to fix the throttling issue however the guide said TS must be in Monitor Only mode to work is this true?

2. I have been using TS turned ON and I'm a little confused how to configure the options so that TS will switch between 2 profiles based on either CPU temp or GPU temp.

Profile 1
Max Turbo if temp is < 85
Profile 2
Throttle down to about x20 until cpu and gpu cool down then switch back to profile 1.

 

This may be of interest to anyone reading this thread:

http://forum.notebookreview.com/asu...5264-bios-cpu-throttling-fix.html#post8196019

 

To automatically switch profiles, go into the Options window and set a CPU alarm at DTS=15. These CPUs are designed to throttle at about 100C so if you want to switch profiles at 85C, the Digital Thermal Sensor (DTS) will be reading about 15. Intel made life confusing by using a sensor that counts down as it gets hotter. If you want to switch profiles at 90C then use (100-90) which is a DTS setting of 10.

I prefer to leave ThrottleStop running and enabled. Some Asus users had success by simply running ThrottleStop which probably disables the BD PROCHOT feature and then they turn ThrottleStop off after that. Whatever works best for you.

Intel CPUs can run reliably right up to the throttling point but slowing it down before it reaches 100C isn't a bad idea for a laptop, especially if it is burning through your lap while trying to use it.

 

For this option, i have one request if you wanna take it for future releases of this fantastic program.

I see this option works great, when the cpu ups to the configured temp it changes the profile, but when the cpu downs the temp, it returns to the previous profile, and the cpu rises again the "max" temperature, and downs the profile again. So, every second changes the profile and the temp goes crazy.

So, here is my request. Maybe, if you configure a timer or a mininium temp for this option to change again to the previous profile, the temp don't goes so crazy and no change the profile every second.

For example, Alarm configured to 15 DTS for change from profile 1 to 2 (first one has turbo enabled and the second disabled). The cpu reaches the temp, program automaticaly change the profile from 1 to 2, the temp goes down. In the case with a timer, maybe in 60 seconds change again to profile 1. In the minium temp case, maybe at 60º can goes again to profile 1.

And if the option of timer/minium temp are configurable too, better twice.

Well, here is my request, i apreciate your work, thanks to that my laptop not reaches the 85º.

Thanks and sorry my english if doesn't undertand very well.

 

Hey there,

I have Thinkpad W520 and I'd like to adjust settings so that I can get the most battery life on days I need it.

Something like lowering the voltage is what I'm looking for. How would I go about lowering the voltage given to my CPU etc. using Throttlestop? I know that not all CPU's are the same and this is why I can reduce voltages to maximize my battery life.

 

We are unable to adjust voltages on the Nehalem/Sandy bridge architectios (any core iX processor). Your best bet would be to lower the multiplier to its lowest setting, turn down your brightness, and disable any unused devices (internal NIC, bluetooth, etc..) I manage to squeezed 5.5 hours of battery life out of alienware doing this.

 

So I have a question, I disabled speedstep in my bios and have the multiplier set to 9.5 but it only shows up as 9.0 in the monitoring window? How do i get it to show 9.5? I also have the cpu setting in power management to 100% for all. I have a Core2Duo P8600 in my laptop.

 

Right click and select "Start dual-IDA". Then run the TS benchmark to put both cores under load. It should show x9.5 for both cores if done right.

 

On an "engineering sample" core2duo cpu that has an unlocked multiplier will TS be able to increase the multiplier by mre than the usual 1x. For example, with an unlocked T7700 will TS show the option of a multiplier higher than 13 (T7700 normal multiplier is 12)?

 

Thanks again nando, you once again saved me. I checked the setting under options restarted throttlestop and it showed 9.5. I reran 3Dmark06 again and got another 400 points with the increase so total is 10815 which is more inline with expected results. I'll update my post in the other section. Thanks.

 

not sure if this got fixed as I'm still on ver 3.0 .... but when I switch from AC to battery and then back to AC, throttlestop would switch the first time but not the second time around. I got two profiles obviously - one for AC and one for battery. And then one for low battery.

so maybe implement throttlestop to check more often the power input status ?

 

brookside: Most ES processors that are not Extreme processors are not unlocked. A T7700 ES is likely going to have a locked multiplier so ThrottleStop won't be able to do very much with it.

miro_gt: If you have found a bug then try the latest version of ThrottleStop to see if that bug has already been fixed. I just tested TS 4.00 with two profiles and went AC - DC - AC about 101 times and I didn't have any problems with it changing profiles correctly.

 

Can TS be used in an OSX86 box, either by running in boot camp or a virtualized environment (i.e., can it be run in win but affect the CPU performance when using OSX)?

Its more than pretty frustrating there is no OSX tool.

 

One user is able to do this and get a higher multiplier in Linux by setting up his Core 2 Extreme multiplier using ThrottleStop in Windows and then rebooting. Whether this is successful depends on if the bios resets the CPU like it is supposed to do. There are some bios versions that do not reset the CPU so the previous settings that ThrottleStop sets might still be available if you boot up into another OS.

Some boot camp users are very happy with TS. It lets them adjust their Mac while running Windows but I don't know if it has any affect if you reboot and run OSX.

 

Some months ago, I reported that explorer.exe was always reading throttlestop.exe from the disk, If i remember correctly, that was fixed.
Now, I noticed that my hdd doesn't suspend, so I decided to see what's going on... and there it is! Windows 7 32bits, explorer.exe is doing queries every second to the throttlestop executable. That causes I/O operations and the hdd never sleeps.
Possibly the same bug is back?

(tools I used: process monitor from sysinternals, free)

 

same problems with v4.00 .

Also, when I start the laptop on battery, it always goes on profile 1, eventhough I got profile 1 only on AC, and profile 2 on battery, and profile 3 on low battery at 5%. Could it be that Dual IDA that is enabled on profile 1 that is giving confusion ?

 

Sadly that's one of the downfalls of using a mac Especially with custom tools like this.. I hope you manage to find a way around it.. Maybe there are people out there who can port the tool - have you tried vmware fusion? not sure if it would give enough access to the CPU to allow throttlestop to do it's magic, but worth a try probably. Unclewebb?

 

Thanks to this tool, I was able to find out why my HP Envy 17 with the i7-720QM has been unusably slow on battery. After a few minutes on battery, the Chipset Clock Modulation gets dialed down to 12.5%. This, combined with a fixed multiplier of 7.0 (which I understand HP does by design) results in an extremely slow system.

My question is, why is my system dialing the processor clock down to 12.5%? Is this a hardware issue? It has behaved this way since it's been brand new. It does seem to be thermally related, as for a few minutes the system runs at 100%, but even under a light load (simple web-browsing) it dials the processor back to 12.5% without fail after a few minutes on battery. When it dials it back to 12.5% the core temps are typically in the mid 70's, which doesn't seem to warrant this aggressive throttling to me. I don't have problems with the system overheating, even after I've used throttlestop to disable the chipset clock modulation.

I've always kept up with the latest BIOS releases and this behavior hasn't changed. I would think that if other people were having the same issue, it would be fixed, as at 7x and 12.5%, the system simply isn't usable.

Any insight anybody could add would be appreciated.

 

I have a P9600 in my HP 8530p and Throttlestop does work in enabling IDA but not when I am docked into my EN488AA HP docking station.

Windows 7 64 bit sp1.

Any ideas why?

 

I don't know why this would happen other than HP seem to be some of the worst system designers out there (from my own experience only) - they are dreadful and will jam it down because it will reduce support costs by a few %. I hate them and will never buy another HP because of their dreadful support.

On a positive note - does Throttlestop help? I mean it should be able to overcome this issue - if so then who cares why - just enjoy your "new" machine AND THANK UNCLEWEBB FOR HIS AWESOME WORK!!!

 

Edit: I found a better way to do things so Explorer.exe doesn't need to query ThrottleStop.exe every second. This and a few other minor fixes will be released shortly in version 4.10.

 

Hi Unclewebb, any chance of including a max performance mode? I was using this wonderful app for months before I noticed how to squeeze the best out of it using the c3 c6 options

Wonderful to hear you're still working on it! You the best

 

ThrottleStop covers a wide range of Core 2 and Core i processors and it has different features depending on the CPU that you are using. That makes it difficult for me to know what is going to work best for each individual user. A lot of the newer laptops lock the register that controls C3/C6. The best thing to do as a user is to sit down and do some benchmark testing like you have done to find out how to get the last bit of performance out of your CPU.

Disabling C1E and the C3/C6 package states can improve SSD 4K write performance by 200% but I have come across very little discussion about this so I assumed that it wasn't that important.

I plan to re-work the BD PROCHOT feature since I recently learned that the bios can reset this after a sleep / resume cycle. Once a few minor issues like that are taken care of, I'll get 4.10 finalized.

Any more bug reports?

 

Hello this is my first post here! I have to say i feel lucky that i found another guy with the same question so easy! From what i've seen from my t7500 ES, the multiplier is actually locked (at 12x), but the good thing is that you don't have to enable dual IDA! My M760tu is going up to 12x with both cores all the time, leading to a T7500 working normally (and being able to downclock and undervolt) at 2.4GHz without any bios mods Next part is to overclock the bus speed a little bit...

The downside of all this is that the TDP is increased, thus i get a lot more than 100oC in Intel stress test and the laptop is running hotter (however the cpu doesn't seem to throttle... could it be bad temp readings?)

 

Is it possible to add support for Conroe-L based Celerons? It's virtually the same chip as Core 2 Duo, having one core and most of L2 cache cut out.

I have built a server based on Celeron 430 and it's passively cooled. I would love to lock it at highest frequency with lowest voltage (1,8GHz @ 1,1V), and I prefer ThrottleStop rather than RMClock.

The biggest surprise is, that this thing supports EIST. For my knowledge, Celerons didn't support EIST until Penryn based ones, and Intel site confirms this. Even though, this one does support it.

More info about CPU: Celeron 430 SL9XN A1-stepping.

According to the Intel site, there wasn't any other sspecs for Celeron 430.

 

I know you can disable some core with msconfig, but would it be possible to choose which one? I've got an i7, and if I disable 4 of them, i've got 2 physical core only with 4 logical, instead of 4 physical and 4 logical. There's no setting in my bios to disable hyperthreading.

Edit: While searching for an answer elsewhere, I heard about "core parking". It seems mine never do.. Also, (right now using 7 core, maybe it has something to do) doing a benchmark with 4 thread doesn't put 4 core at max, but all of them at maybe 70%. Also, cpu=z reports like 2.4Ghz(always high) while TS reports around 1.7Ghz.

 

(crossposting this here as this seems to be the official thread. sorry!)


Hello!

How does Throttlestop work exactly? Does it simply modify the MSR? I'm planning to OC my X9100 from x11.5 to x13.

The reason: I'm running Linux on my laptop. Throttlestop obviously won't run on that OS, but I can read/write MSRs. I tested Throttlestop with a WinXP live-CD and confirmed that it works under Windows.
So all I need to know is what MSRs are being written to, and I can manually recreate Throttlestop's behavior!

I found the MSR payloads in the ini file after saving it. I can write these values to 0x199 just fine. Reducing the FID is possible this way. Rising it above x11.5 is not. Is there any other register that Throttlestop writes to in order to allow FID over stock?
Also, how does forcing the clock modulation work?

Thanks!

EDIT:
No answer, so I reverse-engineered Throttlestop:
The maximum value in 0x198 (0x199 won't accept anything above that) can be changed by writing 1 into bit 16 of 0x194 and the desired payload into bit 0-15 of 0x194.
Clock modulation is disabled by periodically writing 0x00 into 0x19A.

 

Hi everyone,

Just a heads up:

This guide is one of the 18 nominees for: http://forum.notebookreview.com/roundup/641438-nbr-amazing-content-contest.html#post8272122. Again, thanks to Unclewebb for his contribution on NBR.

 

@unclewebb:

First of all: Great Work you've done!

At the moment I use TS only for Dual-IDA-Mode (disableing EIST, C2D P8400). Otherwise I use RMClock. The reason is, mostly I need dynamic power and TS don't support dynamic profiles!

In principle I like TS much more, than RMClock, so it isn't up to date anymore. But I don't like, switching profiles manually.

Is there any posibility, to add a dynamic mode (like 'Performance on Demand' in RMC) in TS???

 

is it possible that throttlestop gives sometimes problems so that when I shut down my computer often I get explorer.exe not responding ? Not every time but happens quite often.

happens on 3 computers that I run throttlestop on.