The ThrottleStop Guide

I'll try to recreate lowing the voltage, if not, this is a problem with all U cpus? with Vaios or how so?
Thanks for all your help

 

This is a problem with all U CPUs from all manufacturers. If you could force the CPU to always stay in TDP Level 2 then these CPUs would run great. The problem is that the bios can over ride ThrottleStop. I think the CPU chipset TDP Level setting takes priority over anything that ThrottleStop lets you do. I do not have any documentation from Intel about this so I have no way to solve it. Intel does not like to share their secrets with me. I do not know of any way to prevent the throttling that you are seeing in your 4500U.

 

Kudos on ThrottleStop 7.00 unclewebb. Works fantastic with the 3920XM and the 4930MX. All new features seem to work flawlessly on the 4930MX. As always, excellent job, my friend.

@jtravapd - I quoted you here. You will find a lot of information about ThrottleStop already posted and can ask more questions from the extremely talented unclewebb that wrote the program.

I will share some ThrottleStop 7.00 screen shots for the 4930MX after work today or this weekend.

 

Hello, unclewebb
Kind of new to these forums.i recently stumbled upon your awsome software while trying to tweak my CPU/GPU(kindof a novice).
Anyways
my laptop spec :
CPU-i5 2450M 2.5Ghz (3.1 Turbo Boost)
GPU-Geforce 630M
Could you perhaps tell me then best settings to run Throttlestop for my Laptop? I do experience FPS drops while playing games such as - Loadout/BF3/Darksiders2. These drops last for like 15 seconds before going back to normal fps and occur often. I think this is due to the throttling issue i have been reading about.Any advise on how to use throttlestop effectively for my laptop. Thanks in advance.



Current Settings so far..i made a log Playing WoW for a few minutes if u would like to see it ?

 

Upload a log file to Pastebin.com - #1 paste tool since 2002! so I can have a look. Are you using the Windows High Performance profile like the docs suggest? Uncheck BD PROCHOT. Your log file while gaming might have some more clues about what methods your laptop manufacturer are using. What laptop model do you have?

Edit: Before you decided to click the 2 different types of Clock Modulation throttling, did you do any testing to see if your computer uses either of these? There is no use correcting problems that don't exist. I have never seen a single laptop that uses both of these types of clock modulation. Many don't use either so there is no need to check these boxes off. It just wastes CPU cycles.

Thanks for your feedback Mr. Fox. Sadly, too many users use ThrottleStop without ever helping with its development. Your screen shots and videos are most appreciated. It's always great when I get to see some of the world's fastest laptops getting a helping hand from ThrottleStop.

 

Did Unclewebb mean people like me that use TS but don't provide feedback? Hehe, I didn't see version 7.00 beta is out.

My specs are as below, and I have talked to you quite a few times before, so I will just get to the point.

AW recently released A05 Bios, which fixed the Processor Current Limit (I believe otherwise known as primary plane current limit). 4900MQ I think gets 85A, and 4930MX gets 95A now, previously both were locked at 55A. The core current limit has been unlocked since A02 or A03.

I am not sure if you have heard of this, but quite a few of us have ran into no POST on the Alienware 17/18. I spent the whole day trying to figure out what caused it, but I can't pinpoint it. At first I thought it's too much of a negative offset voltage for the cores, but look at the screenshot I attached, it's at 1.10V now with 41x, 41x, 40x, 40x. The voltage is from before, where I found that my 4930MX would be stable on 1.10V at 4 GHz.

But no, the no POST is not directly due to that, because I can use your TS and apply it, and the computer boots fine. It's only when I try to apply the offset in XTU and reboot, then it will not POST. Also, it seems like Dell's UEFI BIOS is somewhat wonky. The CPU overclocking option in the BIOS now shows up as "Disabled", and when it's disabled, one can't overclock using XTU, it will always revert back to default values. But with TS, the settings stick. So now I am just using TS, and I have set the XTU service to only start manually, ie when I want it. XTU is wonky, due to its weird interaction with AW's UEFI, and the UEFI is also wonky too, since I have gotten no POST by just enabling the overclocking options. Since A05 BIOS update (or it could be due to earlier updates), the system's default CPU voltage has been lowered to 1.165V from 1.2V, because we have complained to Dell that the default voltage is too high. But what's weird with XTU is that, if I apply default voltage in XTU, TS would see it as 1.165V (to be precise it says 1.167V in TS), but upon reboot, XTU would say 1.2V. And TS reads 1.2V too. There's some conflict between XTU and the latest A05 BIOS.

All one needs to know is, when all else fails, TS works fine.

I noticed that if I don't really have anything opened, during idle the CPU power consumption is roughly the same between "Balanced" and "High Performance", but if you have like 100+ tabs opened, with Steam, a game running in the background, etc etc, seems like with "Balanced" the CPU uses slightly less power, although it's only like a few Watts, like 1-2. In "Balanced", I have seen the CPU idle at less than 0.5W, when the core voltage drops to about 0.6V. But to be honest, although I see slight performance difference in benchmark between the 2 power profiles, in reality I don't really notice any difference.

Apart from the offset voltage which works for Haswell, I can also confirm that Package Power Limits and Primary Plane Power Limits work correctly. It doesn't seem to lock the values if I tick "Lock", but I guess that's probably because of the BIOS. I didn't check on the "Clamp" function but from memory it worked since TS 6.10 I think. All in all, thanks for the added functionality, the control for the offset voltages (which I think I suggested it to you last time), is exactly what I needed.

One suggestion, if TS can have control over the NB frequency (now is ring bus/cache frequency), it'd be great too. Also, having the iGPU clock and voltage control might be useful for some too, although I suspect it will be a tiny amount of people to justify your time and work.

One question, when I set pri plane limits and the power limits, does it update real time, or do I need to restart? I noticed that the offset voltage is applied immediately.

Attached is the screenshot of the settings for TRL. Everything is still subject to further testing.



One last question, it seems like ticking the "Overclock" doesn't do anything for my CPU, is it because it has unlocked multiplier to start with?

 

Oh just to add on, I found out that there's separate offset voltages one can use for different profiles. That's so freaking perfect. I really can overclock my CPU to 4.2GHz or even more now. :thumbsup:

 

Hi, thanks for the reply. I tried to open Pastebin.com but sadly the page is not opening.
I have the Asus K53SM - SX010D laptop. heres the link : 'http://www.flipkart.com/asus-k53sm-sx010d-laptop-2nd-gen-ci5-4gb-750gb-dos-2gb-graph/p/itmd7vs3bhr6jtgb'. maybe this will help? Thanks

pastebin - Throttlestop log 1 - post number 2677290 the pastebin link u asked for..i noticed the fps drops kind of stopped when running throttlestop, but also increased the CPU temperatures..is that correct? is this temp normal?

 

First of all a zillion thanks to unclewebb for this fantastic software. I hv a few questions. 1. Throttlestop shows Tjmax of C2D P7350 to be 90C. Is this correct? 2. Does P7350 not support C3 and C6 states? In Throttlestop they are displayed as dashes. 3. Again PROCHOT is grayed out while BD PROCHOT can be toggled on/off. In Realtemp it does display thermal status as "HOT" when running Intel Burn Test so I guess throttling works as it should... Posted a screenshot of Throttlestop.imagebin.org/301191

 

That's the kind of motivation I need. Thanks.

I was playing with XTU today and watching the register that has the offset voltage info in it and XTU seemed a little buggy. After setting a value and pushing Apply, the offset voltage value didn't seem to get written to the CPU. I will have to do some more testing to confirm this but at the moment I am not surprised that you are having this issue. My plan is to add the Adaptive and Static core voltage feature for the Haswell CPUs to ThrottleStop in the next day or two just so this part of ThrottleStop is a little more complete.

The new Overclock option is useful for the partially locked CPUs like the 4700MQ, 4800MQ and 4900MQ. It is likely that your 4930MX is already fully unlocked in the bios so this feature is not needed. Some other manufacturers might not have the 4930MX fully unlocked in the bios so perhaps this Overclocking feature might be useful for them. If you uncheck this feature and push the Reset Turbo OC button, does this limit your ability to overclock your 4930MX to its default Intel overclock values?

As far as I know, adjusting any of the Power Limits or Clamp feature should change in real time, no reboot needed. You should be able to play with these while the CPU is loaded to get a better understanding of what they do. On my 4700MQ, the PP0 Lock works correctly but when I go into stand by mode, the lock is disabled when I resume. The Package Power Limits on the Y510P are locked in the bios so I can not play with that lock.

k_am1 - I was going to say, your log file looks pretty good. The CPU seems to be running properly now. Throttling slows a CPU down so it consumes less power and produces less heat. If a CPU was throttling and is no longer throttling then yes, it will run hotter. Your core temperature is getting up into the mid 90C range in that log file and once it hit 100C. That's on the outer limits of acceptable. If it was my CPU I would start thinking about pulling off the heatsink and replacing the thermal paste and making sure the heatsink fins are not full of dust.

 

To answer your question, it seems to make no difference whether it's checked or unchecked, I can change the multiplier and offset voltage in both situation. Max multiplier for the 4930MX shows up as 80x. Since using TS 7.00, I have been sticking to adaptive voltage rather than static.

Do you have the Y510P with 4700MQ? What undervolt and what core voltage do you have on it now? I have a profile which limits the multiplier of my 4930MX to 34x 34x 34x 34x, but I can't undervolt it by -100mV, as it will give me BSOD. The most I can go is -90mV. When the CPU is undervolted, TS shows 0.95V for the core voltage. Does this correspond to what you have now?

I have found out the limits of the power that I can pump to the CPU. To be completely stable with the high ambient temp that I have here, I can put at most 85W for long duration power limit. AVX or AVX2 stress test, like the IBT or Intel's AVX2 Linpack, can push the CPU well over 100W, but I am limiting the short duration power limit to 100W for now. Mr. Fox has shown that his 4930MX can go over 130W peak power.

 

Yes, I am using a Y510P with the 4700MQ. I am using the +2 overclock feature so the multis are at 36, 35, 34, 34. At this speed, -50 mV seems stable but if I try to go to -60 mV or -70 mV, I end up with light load BSOD issues. At idle, ThrottleStop reports VID at 1.0365. It can be perfectly stable in Prime95 v27.9 but it's light load where sooner or later I will get a WHEA error BSOD. My guess is that this is C State related. When overclocking on my 3rd Gen 3570K, turning off the Package C States seemed to help with light load BSOD issues but that is not adjustable in the Y510P bios I have so I can not test that.

In theory, a 4930MX core should be a better quality core than the 4700MQ so your MX should be able to run at the same speed with less volts. I am pretty sure Intel has always binned their CPUs so the better cores end up in the boxes with the higher price tag on them.

I am thinking that maybe a static voltage might be the way to go when all of the C States are enabled. At idle the CPU cores are spending 99% of their time in C7 so a fixed voltage might not have any negative consequences since in C7 the voltage is dropping down anyhow. After I do some more programming I plan to look at this further.

 

For maximum undervolt, I noticed that it's best keeping the multiplier all even, like 34x 34x 34x 34x, or 35x 35x 35x 35x. If you have 1 core running at slightly higher multiplier, when under single threaded load, that core will need a higher voltage. Then when you run multithreaded stuff, the 4 cores will run at the higher voltage. The integrated VR is very smart, but still not as smart as a human that can tweak each CPU individually. My 4930MX is barely sipping 20W in Diablo 3, running at 3.4GHz at 0.95V (in TS it's -90mV offset). I think my CPU is actually quite a decent sample, it undervolts very well.

What we really need from Intel, is a dynamic adaptive offset voltage, where it will only apply the offset when the CPU is under load. The adaptive offset voltage in XTU is supposed to do that, but it doesn't.

I don't really have any measurements, but in terms of heat, I find that adaptive voltage will make the processor runs very slightly colder. I see the load and idle temp differ by 1-2C, no more than few Celsius.

I will keep an eye out for the next revision of ThrottleStop.

 

Guys I really need help ..
I don't know why or what is doing it but when I am watching video on youtube it is not playing properly sound is crackling snaping and buzz freezing .. I really don't know why it is doing don't know what problem is .. when am watching normally video downloaded it is fine .. Don't know if cpu is speed enough ..

 

kh90123 - Thanks for your insight about voltages.

GalaxySII - A Core i7-4500U can run very slowly if a manufacturer decides to throttle it. Try running ThrottleStop. It has an On Top feature in the Options window so you can keep an eye on your CPU while watching a YouTube video and you can also check off the Log File option so you have a record of your CPU's performance. There are many reasons for crackling sound but it is usually not the CPU. It would be a good idea to head to the Acer Aspire forum for your laptop and ask questions there. You will hopefully find users that have a better understanding of any problems your laptop model has. It might be a driver issue with the sound chip on your board.

 

Fixed it .. it was Wi-Fi module .. replaced and all is ok ...

 

unclewebb I want to say thank you for updating your program so that it works better with the Haswell CPUs. I am currently running my i7 4700MQ at 3.4 GHZ on all cores on my AW 14 (Haswell 2013 model) thanks to Throttlestop.

However, I am curious and want to do a little test on something since a reputable seller on ebay is willing to sell his i7 4930MX CPU OEM version for $680. Alienware techs at Dell told me that the AW 14 (Haswell 2013 model) can upgrade to the i7 4930MX CPU even though they don't release a configuration with that CPU for the AW 14. In addition, Amazon.UK has a company seller by the name of HIDEVOLUTION that sells the AW 14 with the i7 4930MX and has confirmed to me that it runs well in the tests they have done. They said that they believe it is safe to offer it as an upgrade since they did a lot of stability tests on it. I am about to pull the trigger and buy the i7 4930MX to use on my AW 14 and see if I can reach 4.0 or 4.1 GHZ on all cores using Throttlestop however I am concerned with the heat. I do have modded cooling fans and a nice laptop heat extractor that reduces temps greatly but not so sure. Do you think the AW 14 can handle a slightly overclocked i7 4930MX CPU or do you think it is best if I save my money and just upgrade to the i7 4900MQ? I heard the i7 4900MQ can do 4.2 GHZ on all cores which would probably be already fast enough for video encoding. Can any i7 4900MQ users chime in if they have been able to do 4.2 GHZ on all cores without throttling or not being sustained because of the 47W TDP? It would be great to hear from you i7 4930MX and i7 4900MQ users so I can make a good decision on which CPU I should get to upgrade my AW 14. Thanks in advance.

 

Hello, i have a Thinkpad W520 and used ThrottleStop for long (4.10) when i was playing games or encoding videos. The temperature immediatly dropped almost 10C° lower. I just reinstall windows & all the drivers and i don't know why but now when i turn ThrottleStop on, temperatures remains the same..!! Updating to 6.0 didn't change anything. Here's a capture :



ADD: https://forums.lenovo.com/t5/W-Seri...emperatures/m-p/1013009/highlight/true#M36741 i recall this is the post i saw back in time. I wonder why it doesn't work anymore now. Could it be due to last BIOS update..?

 

Clean your fans maybe?


Sent from my iPhone using Tapatalk

 

Nope, they're clean already I tried to boot on my other disk/older install and i don't have the issue.. The new install contains same thing and driver up to date so i have no clue what's causing this... Hope the dev will know!

 

I went back to former BIOS just to see and no changes.. I'm still thinking there's an option in the BIOS that changed, that can't be explained otherwise, i don't think any driver could cause this, am i wrong?

I tried different things with powermanager but no luck. Still i noticed a strange thing, which is "turbo" and "equilibrate" settings for CPU doesn't do any difference with "turbo maximum" is that normal? Can't recall..

As far as i understand by that, if turning on throttlestop with these option got no impact on temperatures, it means that either cpu didn't enter not in turbo mode, either turbo mode is disabled by something.. Correct me if wrong.

PS: To former advice "clean fan", i was thinking.. well it indeed drop C° to do so, however this is totally off topic here!

 

ddong - After you re-install Windows, there is often times a lot of background activity going on. The screen shot you posted shows that. You can not compare temperatures when this is going on. The results are meaningless. On my laptop when the average C0% at the bottom drops down to about 0.5%, that means my system is idle. Your laptop is nowhere near that level. If your computer is working on something in the background then it is not idle and its temperatures will be higher.

Edit - The CPU temperature sensors are within the CPU so there is nothing in the bios that can change their accuracy. It is possible that you have disabled some of the C States. Have a look at that in ThrottleStop to make sure these are all working the same when comparing. Individual cores spending 99% of the time in C6 or C7 is a good number to shoot for. That's idle.

ThrottleStop 6.00 is better than 4.10 and version 7.00 is better than 6.00. You can use whatever version you want but with each new version I try to include improvements for all of Intel's different processors. I'll spend days going through the code line by line looking for anyway to make things more efficient. This is undoubtedly an OCD disorder but people here seem to appreciate that kind of dedication and constant improvements.

The latest version also handles the Nvidia GPUs better when using Optimus and Nvidia's latest drivers.

It's hard to say if a 4900MQ is going to be significantly faster than a 4700MQ. On paper it is faster in MHz but both of these CPUs are long term limited to 47 Watts. When fully loaded, both CPUs are probably going to be running at a very similar speed. I am not sure what power limit your bios would set for a 4930MX and whether this power limit would is locked or not. If it was locked to the same 47W power limit then once again there would be little to be gained for any apps that fully load the CPU. Your decision depends on if you have some money to burn and if you are feeling lucky. Users on forums can tell you that "it works" but how well it works for the apps you want to run is anyone's guess. Haswell cranks out the heat. This can also quickly become a limiting factor if your heatsink and fan are not designed for a higher TDP CPU.

 

Thank you for the compliment.

Quick question... any idea why ThrottleStop 7.00 would always load maximized at Windows Startup? I have it launch as a task and there is nothing I can identify looking at my settings that would cause it. It bugs me launching maximized instead of to the tray. All of my settings seem to be correct, but I might be missing something. It does this on the M18xR2 and the 18, but prior versions do not. Other than this, I see nothing except excellent results with it. I tried taking ownership and renaming the INI file so a new one would be created, but no luck with that.

Beware of few things. Alienware reps telling you the it can handle it may be inaccurate. Unless they have done this and know from experience, it's probably not a good idea to say anything except " we're not sure, but it would be fun to try it" to a customer. First and foremost, Haswell is an overheating abortion and unless the Alienware 14 has a heat sink specifically engineered to handle the absolutely INSANE temperatures produced by the 4930MX you may end up finding the system almost unusable. It's barely manageable in the Alienware 18 with a heat sink tailor-made for it. Thanks to Intel's moronic design of Haswell, this is a significant challenge and temps are only managed effectively using Liquid Ultra paste. You'd be better off with a humble 4700MQ running at maximum capacity than a 4930MX that is unstable and throttles so badly that it can't keep up with your 4700MQ. Not saying you should not attempt this by any means--I would want nothing less if I had an Alienware 14-- but I can tell you, based on having one, that 4930MX is going to require extraordinary cooling that might not be contemplated by the engineering of your system. It this works well, then I think that would be pretty sweet in such a small package. Hopefully, Alienware will beef up the GPU options. It would likely be a while lot more popular if they built it with a GPU in an MXM slot instead of BGA. If it could run cool enough, an Alienware 14 with 1080p, 4930MX and GTX 880M attached to a 330W AC adapter from the 18" laptops would be one wicked little beast. Speaking of which, the Alienware 14 AC adapter might need to be addressed with a 4930MX as well. What is the capacity of the AC adapter for the 14? The 4930MX can draw about 300% more power than the 4700MQ when overclocked.

 

I can not remember changing anything related to this but I will definitely do some testing on my next days off to try and figure this out. Thanks for the feedback.

Making things more efficient is not always a good thing.

Edit - When run from the Desktop the Start Minimized feature still works. I will try using the Task Scheduler as well. Make sure the Start Minimized option is being saved correctly. Windows does some funny things and tries to block the saving of the settings sometimes. I should cave in some day and stick all the INI settings in the registry and try to keep Windows 8 happy but I would rather not.

Edit 2: I tried using the Task Scheduler and it started minimized OK after I logged off and logged back in. Let me know if you figure anything out.

 

hi thanks this program really cooling my laptop down
but i cant get my maximum performance again i meant MHz i already tried to set back the profile to beginning and try to turn off the throttlestop
what do i need to reset my laptop to default settings?
my laptop is ASUS A43S-M,my maximum perfomance is 2.5MHz and 3,1 when turbo boost activated
now i only can reach 1900MHz,im using ThrottleStop 7.00
thank you

 

Mine start minimized OK. I use Task Scheduler. Basically what I did was just changing the path in the task, to start the new TS instead of old one. Everything else unchanged. I always keep "Start minimized" and "Minimize on close" in TS option ticked.

 

So just so I get this straight, the bios controls whether my TDP is locked at 47W or not? I'm running the A03 BIOS right now on my AW 14. I thought that what determined whether the TDP is locked was the CPU itself. That is why I'm so eager to upgrade to an i7 4930MX because it is unlocked. I know an ebay seller who is selling their M14X R2 with an i7 2920XM overclocked. And according to him the bios allows him to change the multipliers since it is an extreme CPU. Couldn't I just theoretically change the TDP of an i7 4930MX in Throttlestop to 47 W or any lower TDP to make it run cooler? Forgive me if what I'm saying is something not possible.

I'm really willing to take the risks so I can contribute to the all the AW 14 and Throttlestop users out there. Thanks Mr. Fox and Unclewebb. If you have any more advice for me let me know.

 

The AW 14 power supply can draw 150W. Do you think the i7 4930MX would be able to run on the stock AC adapter or no? I'm thinking of just getting the i7 4900MQ and overclocking it to 4.2 GHZ using Throttlestop. Do you think an i7 4900MQ would be stable and sustained at 4.2 GHZ on all cores on the AW 14 under load with good cooling gear?

I'm less inclined now to get the i7 4930MX cause I'm not too sure if my AW 14 would be able to draw enough power for it to run. It is tempting though to want to try. I just don't want to waste cash to find out my AW 14 woudn't run stable with a slightly overclocked i7 4930MX.

 

I actually do have a 330W power supply from my Alienware M17X R2 and it does physically fit into my AW 14 (Haswell 2013 model). So I'm thinking with the 330W power supply from my AW M17X R2 that it will solve all the power draw issues. I'm just worried about plugging in my 330W power supply into my AW 14 for a long term duration. Will this do any damage to my AW 14? I think I'm going to try my luck and try out the i7 4930MX CPU. I just need confirmation if plugging in my 330W power supply will do any long term damage to my AW 14.

 

I found the above info in Volume 1 of the 4th gen mobile processor datasheet.
http://www.intel.com/content/dam/ww...obile-m-h-processor-lines-vol-1-datasheet.pdf

My interpretation of this is that it is up to the bios to set the turbo power limit for the CPU. It is also up to the bios to decide whether to lock this register or leave it unlocked so users can adjust the turbo power limit while in Windows using tools like XTU or ThrottleStop.

The 3rd Gen Datasheet listed specific TDP numbers for the maximum value that the CPU PL1 limit could be set to. The 3rd Gen Quad Core CPUs that had a TDP of 45W (ie. 3720QM) could have the long term turbo power limit set to a maximum of 60 and the short term turbo TDP value could be set as high as 72. The 4th Gen datasheet is not as specific. The public docs no longer list maximum values for the 4700MQ/4800MQ/4900MQ series.

If the bios has not been specifically written for a 4930MX then it is anyone's guess as to what TDP values will be used. I think the bios can still decide to lock the turbo TDP of an unlocked processor, especially if the bios is not sure what the CPU model is. It might pull some default values out of its hat.

47W doesn't go very far in a Haswell CPU. While fully loaded running Prime95 v27.9, long term, the CPU throttles back with an average multiplier of about 28.5 so it maintains itself under the 47W limit. That also includes a little bit of under volting. The 4800MQ and 4900MQ use the same 47W TDP value and will also be limited to this same speed during this test. On the 4700MQ, the TDP would have to be set far higher to maintain the theoretical maximum 34 (32+2 OC) multiplier with 4 cores fully loaded. At this point in my Y510P, it would start to throttle based on thermal throttling kicking in at 99C. Pushing Haswell to the outer limits can be a a no win situation.

The next problem is that Dell used to limit the power adapters in the bios. If your Alienware 14 never shipped with a 330W power adapter then the bios might decide to limit the amount of power from the power adapter. This could starve your system and result in CPU or GPU throttling. I hope you ignore all this advice and go for the big daddy 4930MX. Post a link here so I can go to the AW 14 forum and check out your results.

Edit - The new PP0 Power Limit can be used to limit your CPU to less than the rated TDP if temperature is an issue. This feature also needs to be unlocked by the bios. On my Y510P this is unlocked but on many laptops it is locked.

kamui99 - There is an option in ThrottleStop called Do Not Reset FID / VID on Exit. If this box is NOT checked, when you exit ThrottleStop, it will write the default multiplier value to your CPU so Windows can take over and your CPU will hopefully be able to run at its correct speed.

You can always reboot and do not run ThrottleStop. If your bios is functioning correctly, this should also reset your CPU to its default MHz.

 

My Windows install's new but not that much. It actually stopped "feeling slow" about two days ago. If something's taking up CPU then it's Avira or Firefox. Which have always been opened. So it doesn't explain the actuall issue at all..

PS: About the "turbo boost" when does it occur exactly..? When Windows start, i can see in the Power Manager that CPU's @150% but otherwise, even if running heavy games + bluray playback + more, i do never see the turbo ON..

 

I just did a little test and plugged in my 330W power supply to my AW 14 and it recognizes it as 240W in the bios. Would 240W be good enough to run a slightly overclocked i7 4930MX?

 

ddong - Your picture on the previous page shows that your have Disable Turbo checked in ThrottleStop. That completely disables the Intel Turbo Boost feature.

Clear the check mark from the Disable Turbo box, open up the TS Bench test and select the 1 Thread test and then take a screen shot while the 32M test is running and post that here. That test should show lots and lots of Turbo Boost.

When Turbo Boost is not disabled, it should be working pretty much ALL of the time. CPUs that work quickly when needed and enter the low power C States when they are not busy saves power. That's how Intel designed Turbo Boost to work. The only reason to disable it would be to reduce the peak heat or peak power consumption of your laptop.

Some laptops continuously vary their fan speeds. Other laptops randomly cycle the fan on and off. This can make it very difficult to do any temperature testing. You have to precisely measure and control the fan speed and you have to precisely control the load on your CPU. You can not do this if you have some apps randomly loading your CPU such as Firefox or an antivirus program. If you do your idle temperature testing with the C0% held at a constant number like 0.5% and the CPU fan running at a constant speed then your temperature testing results will be a lot more meaningful. Make sure that your room temperature is the same too.

Hard to say. Are you 3D gaming, how multi-threaded is the game or app, is your GPU overclocked, how hot is your 4930MX running, will it be an ES, does it leak power excessively? Try finding someone that has done this already so they can provide you with some meaningful test results. You might find someone in the Alienware 14 forum but you are not likely to find anyone here with that level of knowledge If you want to try this then go for it.. If it doesn't work out, there is always EBay and that is hardly a loss considering how much fun you will have. Thanks for confirming that Dell still limits their power adapters in the bios.

 

No 3D gaming, but I will be doing a lot of Sony Vegas Pro 12 video editing and rendering. I want to tests its limits when it comes to rendering speed. Your right I could always resell. It is an OEM version by the way.

I'll post a picture of the A03 BIOS confirming that the max power it can support is 240W. Thanks again Unclewebb. I'll keep you all posted.

 

I'm starting to understand it. However when you check "disable turbo" i thought it disable it ONLY when ThrottleStop is on : It's not the case, it's still disabled after you close the app..! Not really normal i think, is it..? Anyways, i totally understand what you said and i agree HOWEVER that STILL doesn't explain this behavior change. Once again, same unit, same windows, same apps, same drivers... And totally different results. Better or worse, depends the apps. I'm just trying to understand why such change...

 

Maybe you checked Disable Turbo and that is why your temperatures dropped 10C. I do not know for sure. I do know that ThrottleStop 4, 5, 6 and 7 all read the exact same temperature sensors in your CPU. If your CPU is truly idle or if it is fully loaded, all versions of ThrottleStop will report the same temperatures for any Core i CPU. There are a lot of variables that need to be precisely controlled when doing any temperature testing in a laptop. Now that you know this, try to do some more testing and post some ThrottleStop screen shots if you see some unusual differences.

Some users like to disable Intel Turbo Boost and when they exit ThrottleStop, they want Turbo Boost to still be disabled. Maybe your idea is best but if I change ThrottleStop to make you happy then other users will not be happy. Do you see my dilemma? Now that you know how the ThrottleStop - Disable Turbo option works, you can plan ahead and make sure this is not checked before exiting ThrottleStop.

 

Here is a little update folks. I overclocked the multipliers of my i7 4700 MQ to 36,35,34,34 and did a BCLK overclock from 99.98 to 103.00 MHz using XTU. This brought my total overclock to about 3.5 GHZ on all cores. Using the 330W adapter, I was able to complete the 32M test in about 8 seconds without any throttling at 3.5 GHZ on all cores. Now when I did the 1024M test, I found that it would throttle when I hit the 12% percent mark to about 2.9 GHZ when I kept the PPO Power Limit to 30. However after boosting the PPO Power Limit to the max 58, I am able to maintain 3.5 GHZ on all cores until about the 18% percent mark of the 1024M test and it throttles only to 3.2 GHZ. The max temperature I hit under max load at 3.5 GHZ was 76 degrees. Just to add, I did a -10 Dynamic Offset undervolt and used my U2 modified cooler with three desktop fans on the bottom.

So in conclusion, the CPU throttling was caused by the TDP locked at 47W and not my temperatures since I only hit 76 degrees. I'm seeing a lesser point now in replacing the i7 4700MQ to an i7 4900MQ since I'm guessing under long term stressful situations at %100 load, the i7 4900MQ (while technically faster on paper) will only hit about 3.2 GHZ on all cores because of the TDP capped at 47W.

I'm still contemplating about the i7 4930MX CPU because of the unlocked TDP. I'm just hoping the AW 14 BIOS will allow me to tinker with that.

 

In the event you do plan on having settings stored in the registry, please make it so that at least there will be an option that can toggle whether TS stores settings in the registry or in an .ini file.

On a different note, I've been away from the forum for some time due to ISP issues and only recently have I got back online. TS 7.00 beta 1 sounds quite interesting and I will do some testing on my system, but I'm puzzled about a few things. For instance, since TS 6.00 already detects my Nvidia GPU (secondary GPU, Intel's IGP runs the main output), temps are already displayed so will there be additional data monitoring available? Also, what are adjustable primary plane power limits? And I assume Turbo Ratio Limits weren't greyed out on locked systems due to BIOS issues, considering the locked systems I've tested TS on all had them disabled.

 

First of all a zillion thanks to u for this fantastic software.I know u r busy with improving Throttlestop for Haswell but I would really appreciate it if you cud spare a few moments to look into my queries plz.I hv a few questions. 1. Throttlestop shows Tjmax of C2D P7350 to be 90C. Is this correct? 2. Does P7350 not support C3 and C6 states? In Throttlestop they are displayed as dashes. 3. Again PROCHOT is grayed out while BD PROCHOT is can be toggled on/off. In Realtemp it does display thermal status as "HOT" when running Intel Burn Test so I guess throttling works as is should... Posted a screenshot of Throttlestop. imagebin.org/301191

 

After a lot of thinking, I decided to pull the trigger and buy the i7 4930MX CPU for my AW 14. I got the QS version for $680 and the reputable seller told me that they have done a lot of tests on the chip and that it performs the same as the OEM version. I've never had any stability problems with QS chips anyway. I'm actually running my AW M17X R2 with a i7 940XM QS chip at 3.8 GHZ all cores stable.

Nevertheless, I know I'm taking a big risk with getting the i7 4930MX as there are so many uncertainties with the heat, if the AW 14 bios will recognize the i7 4930MX and whether my 330W PSU power adapter (only recognizing a max of 240W in bios) will be enough for a slightly overclocked i7 4930MX. I think the power draw should be fine since I'm only running a GTX 765 compared to the AW 18 that runs the SLI GTX 780.

Despite the risks, I won't know if I don't try and experiment. Hoping for the best out of this. I will keep you all posted when I receive the i7 4930MX. It should be coming in between Wednesday and Friday this week. Until then.

 

ThrottleStop 7.00 makes it easier to see when Nvidia Optimus is working.



If the Nvidia GPU is asleep, ThrottleStop will avoid waking it up to sample its temperature. That would be pointless. As soon as you start using any program that uses the Nvidia GPU, the GPU will wake up and start providing temperature data to ThrottleStop. You can test this feature out by starting and exiting GPU-Z. This feature works correctly with the latest Nvidia drivers on my GT755M.

I do plan to add a separate panel of information for Nvidia GPUs. I would like to report the temps of both GPUs in the Y510P SLI systems. Not sure when I will add this feature. Lenovo wants $274.99 for a second GT755M5 for my Y510P in Canada and the exact same card when purchased in the U.S.A. is selling for only $99.99. If Lenovo ever comes to their senses, I will probably buy a second GPU to motivate me to add this new feature.

The adjustable Power Plane Limit can be used to run your CPU at less than its rated TDP. If you bought a CPU with a 47W TDP you could use this feature to convert it to a 37W TDP CPU or you could use this feature and turn your CPU into an ultra low power 15W CPU. This might be useful when running on battery power or if your laptop is burning through your lap. It's a simple way to sacrifice some performance to help keep the heat down. I also added this feature just in case a manufacturer is using this register to deliberately limit the performance of a CPU. As long as this register has been left unlocked by the bios, ThrottleStop can be used to adjust it.

The new overclocking feature and adjustable Turbo Ratio Limits work correctly on my 4700MQ so I can use ThrottleStop 7.00 to unlock the +2 bins of additional Turbo Boost. The 4800MQ has +4 bins available and I think the 4900MQ has +6 bins available.

To be honest, I am never 100% sure if ThrottleStop is graying out the appropriate boxes. ThrottleStop supports a wide variety of Intel's Core 2 and Core i processors but unfortunately I don't have access to enough different systems to thoroughly test everything.

senzazn12 - The TS Bench does not use the newer AVX or AVX2 instructions. If you want to see some serious TDP based throttling, check out Prime 95 v.27.9 or better yet, Prime 95 v.28.1. The AVX2 instructions must be doing something important because they are like Armageddon for a laptop Haswell CPU.

Prime 95 version 28.1
New prime95 for Haswell - [H]ard|Forum

Intel does not fully document TJMax for the Core 2 based CPUs. They show a P7450 with a TJMax of 90C but they don't show anything for the very similar P7350 so your guess is as good as my guess.

http://ark.intel.com/products/36734...50-3M-Cache-2_13-GHz-1066-MHz-FSB?wapkw=p7450

http://ark.intel.com/products/36750...-P7350-3M-Cache-2_00-GHz-1066-MHz-FSB?q=p7350

Regardless of what Intel says, 90C might not be correct for either of these CPUs. It might be 100C or 105C instead. Intel used to pull numbers out of a hat. The temperature sensors in the 45nm Core 2 CPUs were terrible. Intel did not start writing TJ Max information to their CPUs until the Core i CPUs were first released.

The P7350 might support C3 and C6 but ThrottleStop only reports this data for the newer Core i CPUs. The CPU Package PROCHOT feature is also only available on the Core i processors. That register does not exist in the Core 2 CPUs. I believe BD PROCHOT works on the Core 2 CPUs but with limited documentation, don't quote me on that. Sorry, I couldn't see the picture you posted. Try using www.imgur.com

senzazn12 - Congratulations. Fun times ahead.

 

I may have missed something but I don't think I saw an answer to that..?

 

Thanks for taking the time to reply. Posted a new screenshot imgur: the simple image sharer

 

I answered your question already. Many users like how the Disable Turbo feature works so I do not have any plans to change this.

 

Ok, I've done some testing and I have a couple questions as well as some issues I couldn't fix.

First of, what can I do with PP0 Power Limit aside from what you wrote on this post? I've also attached a screenshot of my Turbo Power Limits window for reference. The PP0 Turbo Limit range goes from 0.0010 all the way to 3670016, even though if I set it to high values, it automatically goes back to 28.


Second, I was trying to test how Optimus is recognized but I'm currently having an issue with WWAHost.exe forcing itself to run on the Nvidia GPU, and even after terminating the process the GPU remains active, as indicated by TS. Also, I've had an issue of TS not displaying the tray icons, after which I've ran the notification area cleaner .bat, which partially fixed the issue by displaying the CPU tray icon but the main TS icon still doesn't get visible. Another thing is that the GPU icon can't be enabled through Options as when I check the Show GPU box, the box remains unchecked. This behavior doesn't occur on TS 6.00.

EDIT: Would it be possible to add a GPU load % tray icon to the options? I'm currently using a desktop gadget for that sole purpose.

 

The PP0 power limit was added so you can convert any CPU into a low TDP CPU but it was also included in case any manufacturers are limiting the performance of your CPU using this method. It would be simple for a manufacturer to slow a CPU down significantly using the PP0 register but I haven't heard of any yet that actually are. If you do not need these features then do not put a check mark in the PP0 Power Limit box. The PP0 Power Limit is sort of a subset of the Package Power Limit. Setting it higher or the time limit higher than the Package Limit would probably be pointless. If your maximum Package time limit is 28 seconds then you will not be able to adjust the PP0 time limit beyond that. The PP0 time adjuster will snap back to 28 when you release it.

ThrottleStop has always accessed Nvidia GPU temperature information directly through the Nvidia driver. I am using version 335.23 and I have a GT755M. On this laptop, ThrottleStop lets the GPU use Optimus so the GPU can go to sleep and remain asleep. No GPU temperature information is provided to ThrottleStop when this happens. I do not know how long the Nvidia driver has been handling this correctly. With older drivers, this never used to work. I also do not know what GPU models this works correctly on. When your Nvidia GPU is idle, ThrottleStop should report --. When you start up something that uses the Nvidia GPU, ThrottleStop should start reporting temperature data. When the Nvidia GPU goes back to sleep, the temperature displayed should go back to --. I used GPU-Z for testing purposes and this feature works 100% correctly for me. Some feedback about this new feature would be interesting. The Y510P with SLI will probably continue to report temperature data because using SLI automatically disables Optimus.

ThrottleStop is not fully compatible with Windows 7 or Windows 8, especially the hide the system tray icons feature. The C++ compiler I use was originally released in 1998 so programmers could develop some Windows 98 apps. Obviously, a lot has changed since then, especially how Microsoft handles the system tray icons. Freeware software is good in theory but with no income coming in, there is no money in the budget to upgrade my version of Visual C++. This is not a request for donations, it is simply the facts. It will be a few months but I do plan to have another look at this to see if there is some way I can work around the icon issues. This is the reason I am considering writing a handful of values to the registry.

I created a ThrottleStop folder in C:\Program Files(x86) . After cleaning out the icon cache, ThrottleStop 7 works correctly when run from this folder. If you use 2 different versions of ThrottleStop in different folders on your hard drive, I think this can confuse Windows 7 or Windows 8. When testing new versions I would suggest copying the new ThrottleStop.exe file so that it over writes the previous version you were using.

I think it would be possible to read the Nvidia GPU Load info from the drivers. I will put that on the things to do list too. Static and Adaptive voltages for the 4th Gen Haswell CPUs is at the top of the list.

Edit - You probably realize this but you might need to exit your Nvidia desktop gadget so it doesn't disturb Optimus.

 

The PP0 time limit did indeed snap back to 28 whenever set higher than that. I was just hoping that PP0 could be set beyond the stock's maximum values as well as lowering them below stock.

I should have added that I not only had closed any software that called for the GPU, namely the desktop gadget and a small desktop window color management tool I use to lock the GPU in an active status (given that when nothing's using the GPU, Optimus keeps switching between the Nvidia GPU and the Intel IGP in a cyclic pattern). But after closing both of them, WWAHost.exe kept calling for the GPU, despite it being set to run on the Intel IGP.

The Nvidia 500 series works optimally (so far, at least) with 310.xx-32x.xx driver packages. I've had great stability and performance on 310.90, but the 310.xx drivers were reported to occasionally display erratic behavior, after which I've updated to my current 327.41 drivers. I have heard 335.23 do seem to be stable and with performance on par to the rest of the 33x.xx drivers, but said reports came from Kepler users, not so much from Fermi users.

Basically, WWAHost.exe kept preventing the GPU from going to standby so I couldn't test it properly yet. If you're running Windows 8.1, have you seen WWAHost.exe displayed on the Nvidia GPU Activity tray icon (enabled through Nvidia's control panel) as running on the GPU?

On the Notification Area Icons window, TS 6.00 and TS 7.00b1 are both listed but while 6.00 displays CPU/GPU temps, C0%, CPU clock rate and FID underneath ThrottleStop (title), 7.00b1 just displays "CPU GHz".

TS 6.00


TS 7.00b1


EDIT: If some donation system was in place, I would definitely support it to help get a new Visual Studio license

 

Not sure why it's doing this for me. From a cold boot, reboot or logoff/logon always runs maximized for some reason. My settings are the same as ThrottleStop 6.00. Once minimize to the tray, the TaskBar icon goes away as it should.

 

WWAHost.exe is hard to kill in Windows 8 and Windows 8 SP1 (8.1) and it interferes with SLI being enabled/disabled sometimes. If you have any Metro crApps running in the background you probably cannot kill it. I sometimes can on the Windows Desktop as long as no Metro crApps are running. I don't know if you can disable it in the Modern UI, but I haven't tried since I don't use the Modern UI Start page or the crApps. Even so, I have to change my SLI setting quickly or it comes back again.

 

It does seem to be an initiator of sorts for Modern UI Apps. There are reports of users having high CPU usage a few mins after logon but I personally haven't had that issue. And given I use a local account, boot straight to desktop, do manual updating and don't do any automatic content reporting to Microsoft, I don't believe it might be the issue.

Still, what puzzles me is how the Nvidia control panel disregards the display adapter choice for WWAHost.exe alone. Other software that caused issues at least limited itself to one adapter over the other and locking options, but at least it was consistent.