The ThrottleStop Guide

MaxMulti does NOT work with 3217U - that CPU is a hopeless case of pathetic low-budget Ultrabook garbage.



MaxMulti also does NOT work with 4930XM. CPU multipliers cannot be set with ThrottleStop to a value higher than what they are set at in the BIOS or with XTU.

Yes, it most definitely can but not with a stock vBIOS. I already have it running at the +135 offset max with a stock vBIOS. Plus, that is in an Ultrabook and already runs a bit toasty. It's just not the right type of machine for any kind of meaningful overclocking. The entire chassis with the lid closed is almost the same thickness as just the lid on the Alienware 18 LOL. Not much room for cooling components. This is a drop-dead gorgeous machine, but not really made for overclocking. It does play most games pretty well and the display is insanely awesome.

New XPS 15 9530 QHD+ Touch
Dimensions & Weight
Height: 0.3"-0.7" (8 -18mm) / Width: 14.6" (372mm) / Depth 10" (254mm)
Starting at 4.44lbs (2.01kg)
Specifications
4th Generation Intel® Core™ i7-4702HQ processor (6M Cache, up to 3.2 GHz)
16GB Dual Channel DDR3L 1600MHz (8GBx2)
512GB mSATA Solid State Drive
NVIDIA® GeForce® GT 750M 2GB GDDR5
15.6 inch LED Backlit Touch Display with Truelife and QHD+ resolution (3200 x 1800)
Exterior/Chassis
Machined aluminum lid and palmrest
Carbon fiber composite base
Edge-to-edge Corning® Gorilla® Glass NBT™ display
Silicone palmrest pad
Full size, backlit chiclet keyboard
Glass integrated-button touchpad

I have a different machine with GT 750M and it handles 1.2GHz on core clock without a problem. It's still a rather wimpy GPU, but it overclocks nicely. And, it's WAY BETTER than Haswell Integrated Graphics. It will handle most games nicely with medium quality settings. It's decent for a low-budget entry level gamer GPU.

 

Is it a coincidence that MaxMulti's icon is Dufus' avatar? That goofy face.

Oh, Dufus wrote the program, so it's his own avatar.

 

Thanks Mr.Fox, that's very informative. I did post about the throttling issue on my own Haswell notebook here where you can see when the load is below 47W all cores run at 34x but during load drops to ~30x.



But the only feedback I got even after explaining throttling was not just temperature dependent was

Anyway, I believe this might be the problem


On my notebook this is set to 0, nominal TDP which is 47W for the 4700MQ. IOW the software setting of TDP will be ignored. Another check to add on the to do list for TS maybe?

Since the setting is locked it will probably need a BIOS mod so will have to try that sometime later on and maybe also check the possibility of overclocking the processor over 4GHz. Temperatures may be a problem when overclocked though and the smaller silicon tech does not seem to help as the smaller core mass will IMO heat up and cool down a lot quicker. For instance running Linpack AVX2 on just core 3 at 3.6GHz shows a ~30C difference to core 0. IOW there will likely be higher temperature differential between case and core compared to older architecture, which means a disadvantage with thermal headroom even before the cooling system is taken into account.

That's probably going to be expected for U and Y processors.

Most likely the BIOS is locking the turbo register MSR 0x1AD with the lock bit in MSR 0x194 as explained a few posts back.

 

Thanks! That is awesome information. I had not seen your other post. I appreciate the links.

Very enlightening... it is unfortunate that Haswell sucks like it does. It is sad that so many people had stars in their eyes and were drooling so profusely over something that turned out to be a piece of crap for performance enthusiasts. It would have been nice if the Haswell socket and chipset would have been backward compatible with Ivy Bridge the way Ivy Bridge was with Sandy Bridge. A lot of these new machines would really rock with Ivy Bridge processors.

Question: what notebook is your 4700MQ installed in?

The behavior you are seeing appears that it might be a very widespread problem for Haswell. The response you got was pretty pathetic. It does appear to have a serious power management problem. It does run super hot, but the throttling appears to be power related because it occurs in the absence of (or prior to) overheating. This seems like a design defect to me. I never would have imagined that Intel would produce such a lame product.

 

So I found out that Processor Current Limit in Intel XTU, is Package Current Limit in TS, is Thermal Design Current (TDC) in HWiNFO. Scrolling down in HWiNFO it shows that "Programmable TDP/TDC" is disabled.

Is there a way to modify the TDC? Seems like Dell has locked it down. Adjusting it in XTU and in TS works, but whenever the system reboots it's back to default value again. Any suggestion?

I believe his 4700MQ lives in the MSI GE60.
http://forum.notebookreview.com/msi/722196-official-msi-ge60-2013-owners-lounge-53.html

Maybe his 4700MQ is throttling due to inadequate Processor Current Limit/TDC/Package Current Limit, like what we have on the Alienware.

 

This happens when there's a monopoly or not worthy competitors around. Sigh... I miss the Ghz battle AMD vs. Intel of few years ago. That kind of competition is healthy for a number of reasons and prevents what we are seeing now. I hope AMD will make a roaring come back soon.

 

Out of curiosity but is there any reason why MaxMulti would report that there are no overclock bins available on an i7-4770K? It should be noted that it was set on BIOS to run at 4.0GHz. I did hope that it would work in order for ThrottleStop to have more than 40x multiplier available and usable.

 

"cTDP" (Configurable TDP) is an alternative to "BIOS" overclocking. You use one or the other. It is showing as disable because you have the option for Extreme Edition set to BIOS. I can use this in the M18xR2 with an unlocked BIOS. Unless you have access to the configuration menu in the BIOS is it more or less worthless. Even when it is available, configured and enabled I find it to be of little value. BIOS overclocking works much better for me than cTDP. Using cTDP works OK with a mild overclock, but overall it is a mediocre approach with mediocre results.

Here is what the screen looks like with an unlocked BIOS. Also worth noting, if you launch XTU doing so disables this method of overclocking and you have to go back into the BIOS again to re-enable it.

 

Oh so that's what it means. Programmable TDP means cTDP. Unfortunately in the AW 18's BIOS there's nothing else below the menu once you select cTDP, and I found from testing that cTDP raises the power to the CPU to around 40W from 35 W (overclocking disabled in BIOS).

So it's as good as...doing nothing for the moment. If only I could make the Processor Current Limit/Package Current Limit stick to the value that I want it to be, instead of having it back to default upon restart. I noticed that even putting the computer to sleep will reset it back to default.

 

Yes, it's MSI. Advanced BIOS settings can be shown on the stock BIOS (no modified BIOS needed) to give plenty of current and Watts but to no avail.

MaxMulti does not report available overclock bins, it just sets one of the many Intel registers used to limit the processor. If MSR 0x194 is turbo locked (which MaxMulti does not check) then this too will limit your maximum turbo. It would be possible to report that feature but best if everything is reported from one program IMHO, TS. I think Unclewebb might be working on it but has limited time and resources for it at the moment so hopefully it will come later on.

Interesting, do you know if this has an effect on MSR's 0x648 through 0x64c? Seems some of these might be HW set :/

Yep, that may be what's happening. Thanks for the info.

 

No, I am not sure about the effect on those MSRs.

 

Seems like a lose-lose situation. If you can overcome the TDP problem then there's the high temperatures to watch out for.

Same Linpack test as previous post, that is HT disabled and -100mV offset. GFLOPs are more in line with what I'd expect @3.4GHz with 10000 problem size.


Trouble is running these test for more than a few seconds now results in thermal throttling. The GE I've got just has one heat pipe to the CPU, I think the GT models may have 2 but the way Haswell is I'm not sure if it'll make more than a small advantage.

The TDP throttling was gotten around by changing BIOS IOUT setting. I think this change then possibly misreports the current supplied to the CPU so the estimated energy calculations are lower than true energy but do not know for sure. Seems with my system once the CPU power (not the package power) hits 47W then TDP throttling starts.

 

Hello. I've been trying Throttlestop on my laptop (which has the core i5 - 2410M cpu), however whatever i set the multiplier to, it keeps jumping and it never stays on a stable freq. Any idea what might be wrong? Laptop model is Acer Aspire 3830TG

 

If you want a steady multiplier at idle you would need to disable the low power C States which includes, C1E, C3 and C6. It might be OK to do this on an unlocked desktop CPU but doing this to a locked mobile CPU will increase power consumption and heat so it is not recommended. Disabling the deeper C States will disable Intel Turbo Boost on your CPU which is not recommended.

The other thing to keep in mind is that the amount of Turbo Boost and your multiplier depends on how many cores are in the active state and it also depends on power consumption. During normal use, these two variables can be constantly changing so it's normal for your multiplier to be constantly changing too. Put a load on your CPU and post some screen shots if you are having problems.

Locking your CPU to a fixed frequency might be something that you would like it to do but it's not something your CPU wants to do. On your CPU, a fixed frequency will result in increased heat and reduced performance which is not a good combination for a laptop.

 

If you follow the above method exactly then it should work with or without UAC enabled and whether you are in an Admin or Standard account.

 

Hi guys, I hope you can help me with this question.

I tried using ThrottleStop on my Sony Vaio Duo to prevent it from TDP throttling in tablet mode, and whilst trying out options i pressed "Lock" at the TDP Level Control option in the Power Limits menu. Now my question is, how, if at all possible, do I unlock this option again?

Also, the duo has a i5-4200u processor, which sony apparently decided to throttle extremely aggressively when it is in tablet mode. Does anyone know if ThrottleStop can be used to prevent this? I tried a couple of different settings, but it didn't seem to give any result (yes, I turned on ThrottleStop before testing).

Thanks in advance!

 

Quibbler2412: The Lock for the TDP Level Control feature can only be unlocked if you reboot. On some laptops / tablets, you might have to completely power down including removing the battery.

I don't own a Sony Vaio and I don't have access to a Core i5-4200U so if you want someone to try and answer your question, you need to post screen shots of exactly what ThrottleStop options you tried and what sort of testing you did to confirm that it didn't work. At the very least, I need to see a screen shot of the Turbo Power Limits window.

As Dufus knows, there are other settings that can over ride the ThrottleStop settings. If this is the case, there might not be anything that ThrottleStop can do to help you out. Maybe Dufus has some spare time to write a little MaxTDP program to help out the U and UM owners. These UM CPUs are not nearly as bad when they are allowed to run at their full TDP or a little beyond but they are pretty sluggish when the maximum TDP is forced down to 10 Watts or less.

 

Thanks for the reply. I solved the locking problem by restarting a couple of times and then have the computer turned off for a while.

I've tried to following separately to stop the TDP throttling:

• Checked TDP Level Control and set the value to 0, 1, and 2
• Checked Intel Power Balance and set both CPU and GPU to 31
• Checked Clock Modulation and Chipset Clock Mod
• Set the multiplier manually to 23
• Unchecked EIST, C1E and BD PROCHOT one at a time
• Unchecked the items in C state demotion

Additionally, I restarted ThrottleStop and made sure the settings stuck. I tested the results by opening Guild Wars 2 while in laptop mode. Here the the CPU gladly revs up to a multiplier of 23 and stays there. I then closed the computer to tablet mode. This activates the TDP throttle after a few seconds and the multiplier drops to 8. As you can see from the screenshots below, the max temperature the computer reaches while in tablet mode is only 1 degree higher than while in laptop mode, so I doubt the TDP is a result of more heat.

Here are some images of what ThrottleStop looks like under different circumstances:

Idle
GW2 running in laptop mode
GW2 running in tablet mode

I noticed that the C7 states goes up from 0% to about 15-20% when in tablet mode, but that might just be a result of the throttling.

Finally, here's an image of the Power Limits window, as you requested.


Thanks for taking the time to look at this!

 

Quibbler2412, to try and get a better understanding of what is happening put the attached file in the same directory as TS (with WinRing0) and run it while in tablet mode then post / PM the file created (MSRList.TXT).

 

Hi guys, not meaning to butt in or tread on any toes with this thread but I seem to be sort of running into this TDP issue...

I have a Clevo laptop with a 4700MQ on an HM87 board (nvidia 765m)

when I first received it from a pc specialist and loaded it up (CPU occt small ffts, kombustor burn in test for the nvidia) three cores logged 'hot' in real temp, so hit TJ max and the throttling occured i.e. multi dropped straight to 8x and crippled the cpu speed to 800mhz. Whch was fine, as I realised it was hitting TJmax. After a re paste temps have dropped 25-30c

Now, testing only the CPU it balances the multi at about 27x for four cores at the max sustained TDP of 47w and remains what I consider cool at 75c.

If i then use the graphics burn in test whilst the CPU test is running, the CPU multiplier drops to about 24-25x and the nvidia GPU temp rises from 40c to about 68c

After a few mins the hottest CPU core temp gets into the low 80s and TDP peaks at about 53w then all of a sudden the multi drops to 8x, which seems a bit strange - I would have expected it to just keep dropping the multiplier in increments until it found it's equilibrium?


so, I guess what I'm asking is, do I need to just appreciate that with the CPU and GPU under torture testing, i'm trying to pull too much power from the m/board? and use throttlestop to manually force 4 cores to a lower multi than the standard non turbo speed?

I'm already using the intel XTU tool to drop the voltage 50mV offset from the CPU core and cache to get temps down, I wonder if I should use the throttlestop feature to offset all of the TDP bias towards the CPU cores rather than the on chip graphics?

 

I think if bi-directional prochot is kicking in you might see that, perhaps your regs are drawing too much current or getting too hot. TDP doesn't seem to work in the traditional way with Haswell.

Quibbler2412, I'm off to work tomorrow overseas so probably will not be able to look at your problem until sometime in December. Sorry.

 

Glad I caught you early on.................

 

Hi Dufus! Thanks for the reply. I've run the tool you gave me in both laptop and tablet mode, so you can compare. I've attached the file, as I can't PM you, since I haven't posted enough yet.

Don't worry about when you get back to me. It's not a super urgent issue, it's just annoying (actually, I only discovered this issue because my girlfriend's Sony Vaio Pro 13 does the same thing, only all the time. She might benefit from any fix you might find). I'm just happy that someone is willing to help.

 

Quibbler2412: The pics you posted are very useful to help understand what's going on with your CPU.

The 3rd Gen Core i7-3517UM CPU I used had the same Level 0, Level 1 and Level 2 TDP settings. Each of these levels also has a unique default multiplier so when you exceed the turbo TDP value, Turbo Boost is disabled and the multiplier drops down to the default value. On the Core i7-3517UM, Level 0 had a default multiplier of 19, Level 1 had a default multiplier of 8 and Level 2 had a default multiplier of 24. Here's some more info about the testing I did a few months ago.

http://forum.notebookreview.com/har...31329-throttlestop-guide-169.html#post9345656

For maximum performance, it would be best to lock this CPU into Level 2 which would get you both the highest TDP value for maximum turbo boost as well as the highest default multiplier. What I didn't realize back then is that there are actually 2 different places where you have to tell the processor to go into Level 2. So far, ThrottleStop 6.00 only lets you adjust one of these. We might have to wait for Dufus to return to write up a new utility to fix up the other location that the CPU checks before it sets the Level value. A utility like this could make a lot of U/UM owners happy and a lot of tablet manufacturers not so happy.

Your picture with the multiplier locked to 8 shows the TDP Throttle box is active which means the CPU is throttling due to reaching the TDP limit. I don't think BD PROCHOT is the cause of the throttling you are seeing but it would be a good idea to clear that box when testing.

If you have time for one final test could you boot up, adjust the Set Multiplier value as high as possible which I think is 26T for your CPU, check the Set Multiplier box, clear BD PROCHOT, check EIST and C1E, open up the Power Limits window and set the TDP Level Control to 2 and then click on the Lock box and click on Save after that. Run TS Bench or a similar test to put some load on the CPU and then switch to tablet mode and see if these settings make any difference. It probably won't help but it is always good to see a screen shot just to make sure that ThrottleStop and the Power Limits window are both set up correctly. I don't think either type of Clock Modulation are being used so there is no need to check those boxes. The information and settings in the C States window is also not important.

 

Unclewebb: I've done the testing like you asked, but sadly it didn't change anything. You can see the result here.

Again, testing was done by running GW2, and as you can see, the temperature never reaches anything close to high before TDP throttle kicks in. Still, no TDP throttling happens in laptop mode.

It does indeed sound like the CPU has been set to one of the lower levels when in tablet mode. Setting the level in throttle stop doesn't seem to do anything, but it would be cool to see if setting it somewhere else might help. I hope Dufus is willing to take a closer look at this when he has time.

 

Quibbler2412 - Thanks for your thorough testing. Hopefully Dufus or I can come up with a solution for this someday. When I have the time, I will see if I can figure anything out before Dufus returns to NBR.

 

Hello,

I started used TS quite recently, and noticed something while my GPU is under medium-heavy load. When I set "Intel Power Balance" to 0 (CPU) and 31 (GPU), It reverts to the default after about a minute (0,8). Unfortunately, I cannot change any other settings.



I would prefer the GPU to have highest priority, but it wont save. Heat is not an issue (68 max). Is It possible to lock the Intel Power Balance to my desired setting? I just want my GPU Clock to exceed past the 650MHz throttle. Here is the reading before it resets itself.



My system specs are

Sony SVE15122CXW
Intel i3-3110M
HD 4000 IGP
Windows 8
Locked Bios

 

MinecraftAddict131 - I was playing around with the Power Balance feature back at the end of August but I haven't had much feedback. Thanks for posting those pics.

http://forum.notebookreview.com/har...31329-throttlestop-guide-169.html#post9345656

On the Asus laptop I was testing with, the Power Balance values seemed to stay at whatever I set them to. It looks like on your Sony laptop, these values are being changed dynamically. I will need to add some code to ThrottleStop to check and adjust these values on a regular basis. Send an email to the guy in the About.. box of ThrottleStop if you are interested in testing this once I get it done. I might do some more work on project ThrottleStop next weekend. I am not sure what is possible but I am definitely interested in finding out.

 

unclewebb, can you suggest a few test scenarios to better gauge Power Balance in terms of CPU performance? I ask as I wonder if just putting full load on all threads for 1 minute would suffice, given that it should be Turbo Boost the main feature to be improved with more TDP attributed to the CPU.

 

tried to run v6.00 but it wont let me set my multiplier higher then 11.5
but version 5.00 I can set it to what ever I want, ie 13.5

 

shakennstirred - I just tested the Core 2 overclocking feature on my QX9650 and this feature is still working fine in ThrottleStop 6.00. Did you click on the UNLK button to unlock the maximum multiplier adjuster?

Most of the new ThrottleStop 6.00 features are intended for the newer Core i CPUs. If you can't get the newer version working then you might as well keep using version 5.00.

tribaljet - It should be easy to rework ThrottleStop a little to maintain the selected Power Balance values. MinecraftAddict131 is going to help me test this new feature when it is done. Any game or 3D app should show a big improvement in the Intel GPU speed and in the FPS when the GPU is favored to get most of the power.

 

What about the Package Current Limit that I PM'ed you not long ago? It will save up a value that I want and apply it upon startup right?

 

Maybe. I always forget about all the requests that come in. Just keep spamming my PM or email box until you get what you are looking for.

 

I dare not spam your PM. I shall assume that you might remember it.

The register is currently locked on the AW 18. I found out a way to make the Package Current Limit sticks, by turning on Win 8 Fast Startup. But Fast Startup can be buggy at times, not shutting down the computer when I want it to. It's a shame that Haswell CPUs runs much hotter than Ivybridge CPU at high clock speed, and a shame that Dell is still not doing anything to fix the fan table and the messed up BIOS.

In the Dell Inventory site, where refurbs are sold, more than 90 AW 18 has been returned.

 

Ah, that seems quite a fair share of potential progress.

Thing is, for Optimus systems the Intel IGP's rarely desirable (for mobility I can understand but I personally use it as little as possible) whereas the Nvidia GPU is almost always used first and foremost. With that in mind, allocating a minimal TDP share (say, 25 for CPU and 6 for IGP) ought to improve CPU performance while not increasing temps in the process, correct?
I'm mainly looking for CPU performance improvements, hoping there can be a parallel improvement on the CPU side of things, just like the IGP can have such an improvement.

 

Regarding my request with the Processor Current Limit/Package Current Limit, is it possible to make it so that each one of the profile (Performance, Game, Internet, Battery) be associated with a certain value?

 

Haii Unclewebb,

Can you please help me with this one,My laptop is Lenovo Ideapad Z560,i3-380M,ram is 3 gb,graphics card Nvidia 512 ,310 series.My laptop is getting overheated and the max temp is showing upto 90,what should i do to reduce the heat,temp?i can't play games and use high memory programs,it is getting over heated.I did what was written in this topic,but i can't complete figure out what more to do,i am not an expert in software/hardware,so can you please help .Thanks in advance

 

remobecks: Older laptops can get very dirty inside depending on where they are used. The cooling fins attached to the heatsink can get full of dust and when this happens, the CPU will run very hot. If it reaches the thermal throttling temperature, this can significantly reduce performance.

Sadly, many manufacturers continue to build and sell laptops with barely adequate cooling solutions and that is when the laptops are brand new and clean. I have no experience with your specific laptop model so it would be a good idea to find a Lenovo Ideapad specific forum. If it was my laptop I would pull it apart and I would start by blowing out the heatsink with some compressed air. If that didn't solve the problem then I would remove the heatsink and redo the thermal paste. When a laptop is used as a laptop and is moved around and twisted and flexed, the heatsink can become loose. It's a good idea to check for this if you are having over heating problems.

kh90123: Unfortunately I don't have much time for project ThrottleStop anymore. Individual power limits for the newer CPUs is a great idea. ThrottleStop has this feature for the first gen Core i CPUs but I haven't gotten around to adding it for the newer CPUs and I don't know if I will ever get around to adding that feature. I will start with your first request and see how that goes.

 

Hello, my first post here :hi2:. I am using ThrottleStop to override the non-Dell AC Adapter (in fact a genuine Dell adapter with a failed ID chip) throttling in my Inspiron 9400. It works on the CPU multiplier, but the chipset is still throttled at its low-performance setting 75%... As I happen to have another working genuine Dell AC Adapter, I made these comparisons screenshots:

With Dell AC Adapter.jpg


With Non-Dell AC Adapter.jpg


The change is displayed instantaneously in ThrottleStop as soon as I switch AC adapters. Everything is set at Maximum Performance in the OS.

Is there a way to override the Dell chipset throttling?

 

Petit-Lion - The chipset clock modulation feature has never worked correctly on some of the older chipsets. When you are using your good AC adapter and you adjust the Chipset clock modulation percentage on the left hand side, does that change the values in the Chip column on the right hand side?

Can you post a picture of the CPU-Z Mainboard tab? I will have another look at this but I might not be able to come up with a solution for you. ThrottleStop was never intended to be used on the early Core Duo CPUs but if I can find a simple fix, I will give it a try.

Edit: OK, I spent lots of time doing lots of testing but I didn't accomplish too much. I did a comparison using the Intel® I/O Controller Hub 8 (ICH8) which I think was released in 2007. The desktop computer was using the ICH8-R and the Dell laptop had the ICH8-M. Both were the same revision B0.

http://imageshack.us/a/img32/9053/ui5t.png

The register that ThrottleStop writes data to is called the PROC_CNT - Processor Control Register. The ICH8 documentation I downloaded shows that both the desktop and the mobile chipset use the same PROC_CNT register and it is in the same location. I downloaded the previous generation ICH7 documentation and it used the same PROC_CNT register with all the bits and bytes in the same location. The newer Z77 chipset that I tested also has this same register in the same location so not much has changed in a long time.

Now here's the part that doesn't make any sense. The ThrottleStop code that controls Chipset Clock Modulation works 100% correctly on the desktop board with the ICH8-R but it doesn't work at all on the laptop with the ICH8-M.

When ThrottleStop tries to write information to this register in the ICH8-M, the value in this register doesn't change just as if it was locked. The Intel docs say that only 3 out of the 32 bits in this register can be locked.

Writing data to this register is not totally ignored. When ThrottleStop requests a Chipset clock modulation value of 75%, the ICH8-M ignores that request but turns around and sends that request back to the CPU and instead, the CPU Clock Modulation register in the CPU gets set to the desired value of 75%.

http://imageshack.us/a/img607/4233/264t.png

For comparison, here's the ICH8-R desktop version working correctly.

http://imageshack.us/a/img534/4625/74m7.png

It's possible that this could be a bug in the ICH8-M mobile chip. Perhaps the PROC_CNT register doesn't exist in the ICH8-M so requests to that register are simply redirected to prevent software from going KA-BOOM. I had a good look at this problem but I can't think of anything that I can do differently in ThrottleStop to work around this issue.

 

Unclewebb, thank you for the time you spent investigating. I'm not sure you still need the mainboard tab snapshot, however here it is. The Chipset Clock Mod setting works with both AC adapters, except that the Chip column value is capped at 75% (and the CPU load at 70%) with the non-Dell adapter. (Picture is with good AC adapter). I also checked that ThrottleStop can take my computer (CPU and chipset) to 100% while on battery. The 75% chipset clock cap only occurs when powered by a non-Dell AC adapter, not on battery power.

This non-Dell AC adapter throttle is really puzzling... I also ran Systester (hyper-pi equivalent) under Linux Kubuntu while playing with cpufreq-utils with a similar result: cpufreq-set works with a Dell AC adapter or with battery, cpufreq-set has no effect with a non-Dell AC power.

 

Petit-Lion - You have the previous gen ICH7-M Controller Hub. The Intel documentation shows that the chipset clock modulation stuff that ThrottleStop does is the same for the ICH7, ICH8 or anything newer for that matter. The ICH8-M in my Dell laptop seems to have something wrong with it. At least the Chipset Clock Mod is working correctly for you.

Can you do a few more tests? With your good adapter, can you run two SuperPi 1M benchmarks. Set Hyper Pi to use only 1 processor. I am interested in seeing what times you get when Chipset Clock Modulation is set at 100.0% on the left hand side of ThrottleStop. Then I need to see the same test with Chipset Clock Mod set to 75.0%. Next, switch to your non-Dell adapter and run those 2 tests again. I am just trying to see if ThrottleStop has any effect. ThrottleStop might be setting that register to 100% when running on the non-Dell adapter but it might be getting changed back to 75% almost immediately. If I increase the frequency that ThrottleStop monitors and adjusts this register, maybe your performance can be improved with the non-Dell adapter.

 

Hi guys!

I'm running a 'gaming laptop' with an i7-2670QM, GTX 560m and 16GB DDR3 RAM and I'd been having some crappy FPS issues with Battlefield 4 - even on low settings. I eventually overclocked my card some and noticed an increase, but a FPS spike every now and again. When looking at MSI Afterburner, it showed the GPU usage would bottom out every few seconds before going back up.

Some research led me here, and I unticked BD PROCHOT in Throttlestop - instantly, my FPS became smoother than I've ever had it...although the CPU temps for all cores now seem to be somewhere between 38-45c idle and around 80-81c when playing the game. Oddly, my GPU maxes out at around 75c and isn't an issue.

I'm concerned that this might be a little dangerous, but obviously the FPS increase and smoothness is absolutely superb.

So...

1) Are these temperatures going to make my CPU die in a blaze of glory soon?
2) Is there anything I can do to lower the temperatures (not thermal paste...I can't open this thing, it's awfully designed)?

 

Intel is a technology company and they are in business to make money. If running their CPUs at a high core temperature was costing them money because of too many warranty claims, it would be simple enough for Intel to lower the thermal throttling temperature to whatever temperature was necessary to prevent damage.

Intel set the thermal throttling temperature for their 2nd Generation mobile CPUs to 100C and the 3rd Generation CPUs were set even higher at 105C.

ARK | Intel® Core

ARK | Intel® Core

This is considered to be the "maximum safe operating temperature". This is obviously a company that believes in their CPUs ability to operate reliably at temperatures that look a little scary to the average user. Intel is not too worried so I don't think you should worry either.

I gave my 3rd Gen desktop CPU a thorough work out and both the CPU and on die GPU ran reliably, fully loaded, at over 100C.

http://img27.imageshack.us/img27/6216/torturetest.png

After adding a better after market heatsink and fan, this CPU runs even faster.

If ThrottleStop was causing CPUs to catch on fire or burn out, I think there would be quite a few reports in this forum telling people not to use ThrottleStop. After years of development, unhappy ThrottleStop users are hard to find.

You can spend some money on a cooling pad if you want to try and run your CPU a little cooler but why bother. As long as your CPU is not reaching the thermal throttling temperature then it is running within the Intel spec so it should last a long, long time.

 

Thanks for the reassurance, dude! Weirdly, a few hours after posting the question, my CPU's core temp came right down to a max of around 75 - which was the same temp as before taking off the PROCHOT option!

Seriously, you have NO idea how much happier it's made me to have Battlefield 4 and every other game I currently own running smoothly on high settings - it really takes the edge off my law degree! - thanks for making a genuinely super program!

 

Hi Unclewebb, here are the Hyperpi tests you asked for:

Dell Adapter - Chip 100%.jpg


Dell Adapter - Chip 75%.jpg


Non-Dell Adapter - Chip 100%.jpg


Non-Dell Adapter - Chip 75%.jpg


Another test you could do and maybe get a clue from: Start your Dell laptop and enter the BIOS. Go to the page which shows the AC adapter ID "Dell AC power adapter 90W" or something like that. Unplug your AC adapter, it shows "Battery power". Plug it back. You'll see that the power supply status is refreshed instantly in the BIOS screen. If you had an non-Dell AC adapter, you'd see that when plugging it back, it takes several seconds before the BIOS displays "Unrecognized AC adapter" or something like that. Seems like the polling routine timeout. This polling routine runs only once as soon as you plug an adapter. If the ID signal wire in the adapter DC-side cord went broke while the laptop being AC-powered, it would be ignored and the laptop would continue to max performance and the battery would charge until next start. This allows a kind of workaround like "plug it forcefully for a few seconds and release when the battery starts to charge" when the failure is caused by a loosy motherboard connector.

Hope this helps, thank you for your great work!

 

I have this problem because I'm not sure if my CPU is throttling or not but it annoys me so much when I play CS:GO or other game and the screen freezes for 1-2 sec every 15-30 sec. I can't play normally. I have a laptop Asus G750JX with an i7 4700HQ processor. What I've done is downloaded the program and set it up using some guides from YT and this forum. This is how it looks like.


Can you tell me what should I do to get rid of the FPS drops because it annoys like hell, especially while playing FPS games with your friends and you get killed within a freeze.

 

MrLazio - If your game is freezing then the problem might be the graphics card or the graphics card driver that you are using.

I would run ThrottleStop in monitoring mode and turn on the Log File option so you have a detailed record of your CPU's performance. Keep track of what time you have these freeze ups and then when you are finished gaming, exit ThrottleStop and Copy and Paste the log file data to Pastebin.com - #1 paste tool since 2002! or somewhere convenient and then post a link here so I can have a look.

Petit-Lion - Thanks for posting that data. Unfortunately, ThrottleStop is not doing anything to improve performance when you are using your non-Dell power adapter. I don't think there is anything I can do in ThrottleStop that is going to change that.

 

Hi Unclewebb, I'm sure you've come across this issue, maybe it's an intended feature in win7 rather than a 'qwerk'

if I let the laptop sleep, on resume, or if I pull the power cord whilst using the laptop, I notice in real temp that the cpu speed becomes fixed to 15 x 99.77mhz.

If i then start throttlestop, even in monitoring mode, it seems to give something a kick which then re enables the full speed

I've been using the balanced power plan fixed at 100% min and max and active fan control for both battery and ac

I did follow the instructions to add throttlestop as a scheduled task (and have used task scheduler before, use it for real temp etc.) but it seemed to disable throttlestop from being run at all, even when I removed the task and rebooted. I had to delete the whole folder and redownload throttlestop so I'm shying away from it. additionally it seemed to crash multiple other startup processes. It's probably my mongrel win7 install but it's stable enough otherwise...

 

RealTemp and ThrottleStop both use and depend on the same WinRing0 driver. Starting both programs at the exact same time using the Task Scheduler seems to cause some sort of Windows conflict. Last time I tested this, I could only get 1 of the 2 programs to start reliably.

I would suggest starting one or the other but not both at the same time using the Task Scheduler. Another option is to delay either program so one starts up immediately and the other starts up 30 seconds later. I think ThrottleStop has a delayed start INI option to give you some finer control over the delay period.

DelayedStart=

I haven't tried using this in recent memory so hopefully this option still exists. For a 10 second delay you would set this to 10. This might give Windows enough time to get its act together when booting up. A supposedly multi-tasking operating system that still seems to contain some code from an era long ago before multi-core consumer processors existed.

You could also try the Task Scheduler delayed start option for this.

Give it another try and let me know if you can get this sorted out.