The ThrottleStop Guide

Yes, C2D seems like a long time ago and memory wanders, you need C-States enabled for dual-IDA unless TS8 now forces one core into an idle state to be able to trigger it.

 

please explain?
what else is needed to get that somewhat close with my i7-4810mq?
also what are your oc settings like?

impressive indeed!

 

He did a water cooling mod, you would need watercooling to reach anywhere near that on your i7 4810mq.

 

@MigaBoo It may be possible to unlock your processor however you already can clock to 4.2GHz just by using the normal limited unlocked multi's and a bit more than that with some extra bclk without having to modify BIOS firmware. Expect a ~10C increase at 4.2GHz with TSBench if scaling is okay. Your 4810 should be better binned and with adequate power delivery and cooling should be able to perform better than my 4700MQ. Hardware plays a big role so it may not seem like that if the HW and/or settings are poor.

My own OC was set for fixed voltage, with 44x core and 43x cache, nothing serious at this time. Your own settings are for offset which when using stepped multi's and limited unlocked bins is not ideal, better IMO if you use interpolation instead.

Whether you can run 42x on 4 cores without firmware modification will depend on your microcode update version. What version does your BIOS setup report and what version does Windows report?

Also for stability testing try using default 38x on all cores with increasing negative offset to start with to find your stable negative offset value. After that the limited unlocked bins can be tested for stability using the same negative offset but this time using the core voltage setting to provide extra voltage for those extra bins. Start off with 30mV per extra bin and test for stability while reducing core voltage.

 

Hi to all.
Thank you for your great software Unclewebb.
I am using it about 6 months with Lenovo Y50 (4720HQ). I set the cpu cache/core and intel gpu to -85mv. And after that cpu maintains the 3.3Ghz speed under full load. As you may know 4720Hqs' freaquency is 3.6Ghz with HT. There are a lot of new great settings in beta 5. Which settings should I set to get a 3.4Ghz I know it is impossible for 3.6Ghz.
Kind regards.

 

@methyn - You mention that your CPU presently maintains 3.3 GHz under full load. What kind of load? Why not include some pics or log files from ThrottleStop while you are testing? If you are running a stress testing program like Prime95, a 4720HQ will be limited by the 47 Watt TDP limit so even with ThrottleStop, it will throttle by reducing the multiplier so it does not exceed 47 Watts. I do not think there has been much success getting beyond that limit when using a HQ processor. If your multiplier is topping out at 33 during a stress test then it is likely being throttled by the 47 Watt TDP limit.

In the ThrottleStop - FIVR window, is there an overclocking option available or is it locked? The default multis for a 4720HQ are

1 core active - 36
2 cores active - 35
3 cores active - 34
4 cores active - 34

The 4720HQ might support +2 bins of overclocking so if that is available, the maximum multipliers would become,

1 core active - 38
2 cores active - 37
3 cores active - 36
4 cores active - 36

Those are the maximums. If you are running a stress testing program that is trying to exceed the 47 Watt TDP limit then your CPU will throttle down to the same speed whether you are overclocking or not. If you use software or play games that are mostly single threaded, the CPU will spend most of its time with only 1 or 2 cores active so in that situation, overclocking might make a small difference to performance. A 38 multiplier vs a 36 multiplier is a difference of only +5.6% (38/36) so most people are not going to notice that during normal use. When overclocking, you will probably have to reduce the amount of negative offset voltage to maintain stability which can increase power consumption which can increase throttling if it causes you to reach the 47 Watt TDP limit sooner.

Bumping up the VCCIN can trick some of these CPUs so they report that they are consuming less power. This might get you a little more turbo boost and a higher multiplier when fully loaded but adjusting VCCIN is more practical for someone going after world records than for 24/7 use. It is not worth cooking an HQ CPU since these things are soldered to the mainboard which would be an expensive repair.

Stuff like that needs to be repeated.
You are welcome.

 

'thanks!

meaning, like raise it to 42 and reduce my core and mv offset in increments?

also why does my core voltage drop from 1.164 to 0.730 to 0.880 etc. than goes back to normale?
never seen this before


edit:
so i "re-did" my oc without touching voltage at all, and even got a slight better result and am able to keep a high cache! i tried at all 40x and 42x without changing voltage, it was only stable for an hour.

 

I finally managed to do this. Here are some screenshots. The temperature did go up to 60 though with only the browser open (no videos). And that even with my cover off. I keep it permanently off since I only use it as a desktop nowadays and the heat dissipates much faster this way.

Before disabling SpeedStep:


After disabling SpeedStep:

 

Congratulations @Artaios. You have discovered Dual IDA mode.

Intel says the P8700 can run reliably up to 105°C so no worries at 60°C.

http://ark.intel.com/products/37006...-P8700-3M-Cache-2_53-GHz-1066-MHz-FSB?q=p8700

If you would like to decrease your temperature a little, the best thing to do would be to decrease your core voltage. You should be able to run the 10 multiplier reliably with the VID set to somewhere around 1.075 V. To change the VID, you need to first enable SpeedStep, make a change to VID and then disable SpeedStep and ThrottleStop will show the updated VID voltage. Less voltage equals less power consumption and heat.

On these CPUs, I like running Prime95 for some quick stability testing. I would drop the VID 0.025 V, do some Prime95 testing and if it doesn't crash, drop it down some more. Keep doing this until you get an idea of how much voltage your CPU needs to run reliably. During initial testing, I only do about 5 minutes of Prime95 testing at a time. Once I start getting closer to an acceptable voltage, then I test for a longer period of time.

@Dufus - I have never had any trouble getting my T8100 into Dual IDA mode with the ThrottleStop - C States disabled. I know the non-K Core i CPUs need at least C3 enabled for full Turbo Boost to work correctly but that doesn't seem to apply to these Core 2 Duos.

 

Hi Guys,

I've read the last 50 pages or so trying to get informations on how to force turbo boost or keep the TDP at 25W on my skylake i7-6650u .
Did any of you made any progress ? PL1 keep kicking and bring back the TDP to 15W after few seconds no matter what settings I apply in Throttlestop or XTU...

Would really love to be able to unleash the power of this processor more than 30 seconds...

NonosBook.

 

Hi guys, I have a Gigabyte P35x v3, i7 4710hq and gtx980m sharing two measly heatpipes and two fans. Without any modifications, it was throttling very badly, with temps reaching 95oC on the CPU and 91-93oC on the GPU. Repasting lowered temps, but it would still throttle with the same temps, it would just take longer to get there.

I disabled TurboBoost in the power settings (lowering the max CPU performance to 95%) and that eliminated the throttling. Max temps were 89-90oC on the CPU and 85oC on the GPU.

Recently I started running ThrottleStop instead of disabling TurboBoost in power settings. Now I tried reading as much as I can about correct settings, please let me know if there is anything I can change. Under FIVR, I set the CPU Core to -0.0508V and CPU Cache to -0.0508V, Turbo Ratio Limits are all locked at 25 (stock speed, no Turbo Boost). Pushing the voltage offset to more than -0.0508 BSODs after playing FO4, although with Prime95 I can push further. Everything else is default.

Any other settings I can modify to lower temps? The laptop simply does not have adequate cooling, and I accepted that. I am fine with no TurboBoost, but I would like to be able to get a couple of more degrees cooling.

 

Repaste your CPU+GPU with some Gelid GC Extreme... Replace the thermal pads on your laptop with some Phobya XT 7W/mK pads which are pretty cheap and you should be able to play with Turbo Boost etc,...

 

Hey unclewebb, thanks for the awesome program.

My old Dell Studio 14z (1440) seemed to be suffering a lot more than it should with regular tasks and only now did I realized it was not running at the T6600's 2.2Ghz under Windows 10. It only got up to 1.6Ghz as reported by the task manager.

As soon as I started TS8 I feel like it got up to 2Ghz max, but still .2 Ghz short of its max speed. In TS8 it would show the max 2.2Ghz, but not in the task manager.

It seems that the chipset clock modulation option is not sticking/working, as you can see from the screenshot below. The laptop uses a nForce 730i motherboard, and I did try to install the latest chipset drivers I could find already to no avail.



Any ideas or hints?
Thanks!

 

You are welcome.

In your screenshot, your T6600 is running at 2200 MHz but as you noticed, it is using Chipset Clock Modulation throttling. The Chip column in the monitoring table is showing 87.5% when it should be showing 100.0%. The Task Manager and CPU-Z probably combine both types of clock modulation throttling with the CPU MHz and then they come up with an equivalent value of approximately 2000 MHz.

The Chipset Clock Modulation feature is only intended for motherboards that are using an Intel chipset. I am not surprised that this is not working correctly on your nForce chipset. On the left side if you set this to 50.0%, what value shows up in the monitoring table on the right side? Does it always show 87.5% regardless of load?

Try running ThrottleStop after booting up. Delete the ThrottleStop.INI configuration file before you start ThrottleStop so it will only be in monitoring mode. Any difference or still 87.5%.

The nForce 730 chipset is almost 8 years old. Without an nForce based laptop for testing purposes and without any documentation, I do not think I can solve this problem but perhaps your testing will give me some more information.

 

No, meaning start with all cores at 38x max and find your maximum stable negative offset.

For the extra bins up to 42x leave the negative offset to what it was set with 38x and just adjust core voltage.

For example say -60mV was our stable negative offset at 38x and while the CPU is cool (idle) at 38x the VID (Vcore) is 1.050V and we want to run 42x. Allowing 30mV per extra bin to start with we want an additional (42-38) x 0.030V = 120mV (0.120V). Now take the idle 38x VID of 1.050V and subtract the offset, 1.050 - -60mV = 1.110V then add the extra bin voltage, 1.110 + 0.120 = 1.230V. So we set core voltage to 1.230V which will give a vcore of somewhere around 1.170V at 42x depending on temperature and load. Then reduce the 1.230V to find the minimum stable voltage at 42x.

Possibly the monitoring software is waking an idle core and reading the VID from there although the CPU itself will be using the highest request of all cores.

 

Delete . Double post

 

You are absolutely right. I just did the test with my i7 surface pro 4.
With an ice pack on the upper right corner (when facing the surface).
It will run at 25W forever. (has been 30 minutes now). I am at full speed 3.4Ghz, with an undervolt (done with XTU) of -90.
With the ice pack it will never go over 78-79 degrees.

So this is clearly not a design limit or real power limit but software limited.
In my case once 86 degrees is reached it will immediately throttle down to 1.2ghz until temp reach 52 degrees.
So the big question is how to be able to raise the temperature limit (before the throttle) closer to the design thermal limit.

Any comments ?

 

@NoNosBook - Some tablets and small laptops are using skin temperature sensors to trigger throttling. Try running ThrottleStop and make sure BD PROCHOT is disabled. Some throttling schemes can be fixed by doing that.

 

Hello,

Thank you for your message.
Unfortunately that did not help.
I did disable speedstep to allow BD PROCHOT to be disable.
As soon as the processor is reaching 85-86 degrees. The sensor on the upper right of the surface do trigger the PL1 to kick in and drop the TDP to low values.

Otherway to let the sensor go over this "limit". According to manufacturer specs the processor can handle 100 degrees. So there is no reason to limit it at 85-86 degrees.

 

Leave SpeedStep enabled when using ThrottleStop. You can disable BD PROCHOT without having to disable SpeedStep. If disabling BD PROCHOT does not solve your problem then there are not many other options available.

The U CPUs have 2 or 3 different power levels. The Default power level and usually a high power level and a low power level. The ThrottleStop Turbo Power Limits window should show specs for your CPU. The low power level has a very low maximum multiplier associated with it. Once a CPU is forced into this TDP level, there is nothing you can do. It will be left running like a slug.

It is not the CPU that they are trying to protect. Microsoft's lawyers are trying to protect against lawsuits due to the surface temperature of their devices getting too hot. The new skin temperature sensor is Intel's solution. It might be a nice idea for your grandma but it is a pain in the butt for enthusiasts. A modified bios might let you access the temperature limit for this sensor.

 

Thank your for your message.
I agree with you, without a bios mod it might be impossible to trick the sensor.

However, i do think there is a workaround.
We know fore sure that the PL1 is trigered when the temperature reach 85-86 degrees and it will immediately lower the TDP drastically and bring down the temp almost instantly to 50-60 degrees.
What if we could dynamically adjust the processor max frequency and stress test for 30 minutes so it will reach 83/84 degrees but never more. It might allow to be close to maximum TDP , surely not 25 W but maybe something like 21-22W.
I ve tried adjusting the multipliers in throttlestop without any results (it simply stays at 34) and I do not have the option in XTU.
No way to modify multipliers if it is not a K processor ?

 

ThrottleStop 8.00 Beta 6
https://www.sendspace.com/file/of4xwj

December 29, 2015

New Features:
- enabled overclocking and voltage control for 6th Gen Skylake CPUs.
- added FIVR Faults and FIVR Efficiency adjustment for Haswell and Broadwell CPUs.



I noticed when using Intel XTU that the FIVR Faults and FIVR Efficiency settings were not being restored correctly after resuming from stand by so I added these to ThrottleStop. When the FIVR Faults are disabled, you can enable them in ThrottleStop and then you will have to go through a Stand By / Resume cycle to reset the CPU or you can reboot. XTU forces you to reboot when doing this.

Edit - Here is a simple explanation from an AsRock Z87 manual.

Full Skylake support is still a work in progress until I can afford a new Skylake system. Maybe next year.

 

In the FIVR window, what is the Non Turbo Ratio set to? Is it locked? If it is not locked try setting that to 0 and make sure SpeedStep is enabled. Does that give you any control of the multipliers?

In the Turbo Power Limits, is your Package Power Long limit adjustable? If it is, try setting that a little lower to avoid the trigger temperature. You could also try using the PP0 Power Limit if the Package Power Limit is locked.

 

Hey,

Great work as always! Thanks for making such a great program. When is the beta 6 set to expire, btw?

Thanks again.

 

Hi @unclewebb

Firstly thanks for your continued development of TS. It's a brilliant program and I use it 24/7.

With Skylake support being a work in progress and you not having a system of your own to test on could the community help you in any way?
I have a Skylake machine and would love to help if I can with my 6700K.

 

First of all thank you for your continuous support and the release of the new version.

I have been able to unlock and modify the multiplier and adjust TDP (it is not locked either).

But the results are very disappointing...
If TDP is set 25W PL1 will kick in at 86 degrees
If TDP is set 22W PL1 will kick in at 81 degrees
If TDP is set 20W PL1 will kick in at 78 degrees
If TDP is set 19W PL1 will kick in at 74 degrees
...

So it means that the sensor has been configured to kick PL1 with a ratio TDP/Temp. It is even worst...

Does that mean the only solution is a bios hack to reconfigure the sensor? and that is pretty clear it is not going to happen anytime soon.

 

im using a laptop asus n551jm with i7 4710HQ. I have noticed a very interesting behaviour regarding PROCHOT, in my case Asus has limited it down to 85 C, but, occasionaly, on some reboots or wakeups it sets itself to 95 C. Simply rebooting doesnt usually do the trick, more often i have to go into BIOS and reboot from there, sometimes it works after booting when laptop was off for some time. Although none of the above actions are guaranteed to set PROCHOT to 95 C, i wish there was a more sophisticated method

 

Since it's an external sensor, it has to report to the EC, which in turn is throttling the cpu.

Unlike Intel's embedded sensor, this thing should be out in the open and, once found, can be relocated to somewhere cooler. Look for two small wires joined at the end and attached with glue to a metallic part of the shell:

Spoiler: temperature sensor

That's a thermistor, ubiquitous in all kinds of temperature measurement. The cpu itself, too, has one, except in micro smd form factor. Anyway; find it, dislodge from surface and place somewhere cool. Could have it dangling in the air, perhaps. Or clip its wires and extend them in order to reroute it through the display hinge and attach it somewhere inside the display lid. That's probably the coolest part of the laptop.

 

Hello and happy new year !

I've watched all available videos (ifixit and others) about the complete teardown of the surface pro 4. Unfortunately, I cannot see a thermistor anywhere. Moreover, tearing this is a big mess and moving the sensor to somewhere else will be very complicated as everything is very packed.

I guess the only solution is to wait for bios hack to be able to tell the sensor not to kick in PL1.

 

Hello and happy new year !

I've watched all available videos (ifixit and others) about the complete teardown of the surface pro 4. Unfortunately, I cannot see a thermistor anywhere. Moreover, tearing this is a big mess and moving the sensor to somewhere else will be very complicated as everything is very packed.

I guess the only solution is to wait for bios hack to be able to tell the sensor not to kick in PL1.

 

What you are witnessing is ASUS being ASUS. The CPUs have temp limits between 95c and 100c, but ASUS likes to do things like... set CPU throttle point on already-hot CPUs to 85c. Set temp throttle points on GPUs to about 75c from the default 91c.

Don't buy ASUS in the future =D.

 

@unclewebb - still testing, but so far this works great with the 6700K in the Sky X9. As usual, awesome work. If I find any bugs I will let you know. I have not yet and very pleased with it. Thank you for making this for us, bro. I can hardly wait to see what "full Skylake support" looks like when you are done with it. Sweet!

 

Yes thank you! I cant wait to try it out!

 

unclewebb, happy new yr man, just dropping by for a quick question. my 940xm and 920xm can't seem to adjust vid for some reason, area is grayed out could this be limited by the system bios thats affecting throttlestop?

im using both respective cpu in w701ds from lenovo, both have VID portion grayed out, using throttlestop 6.00. i can't use xtu on any of the first gen i7 dono the reason to it just never works for me, and lenovo bios have nothing.

 

Two years later, Happy New Year again to all and especially to @unclewebb. I've been using TS all the time without a single problem. I have two quick questions again:

- CStateLimit changed again to 511. I didn't change it, as far as I know. What I noticed is that it is now MAX for the first two profiles in the PACKAGE C STATE LIMIT box, and C1 and C2 in the last two prrofiles. The box is grayed out so I cannot change it through the GUI, but I guess could edit the value, and I guess the setting is intentional. Can I set it again to FFFF (or decimal)?

- Is there any new worthy setting in the new version(s) of TS for my trusty old desktop Sandy 2600K?

 

Hi I have some questions regarding to the TDP setting and throttling policy of the Broadwell Core M platform. I found that Core M has a rather different TDP limit scheme. For example, the default TDP limit is set as 4.5W, but the reading of TDP limit in TS shows both short boost and boost power limit as 15W (I guess these values are obtained from 0x610 MSR). In contrast, the XTU tool correctly shows the power limit is 4.5w (when I modify the power limit in XTU, the 0x610 MSR doesn't change).
As a result, the XTU could let the CPU work with a higher TDP while when using the TS, the TDP will still be throttled back to 4.5w even the 0x610 MSR is set to a higher value (limit reason shows PL1 limit). So my guess is that Core M has somewhere else to hold the long term TDP constraint but I don't know where it could be (I did the diff in MSR and check the Intel developer manual, neither of them give a clue) Could you shed some light on this issue?
Another question is related to SpeedStep. Although by using XTU, the Core M can work at a higher TDP limit, sometimes my machines will encounter very weird throttling (the power will suddenly be throttled back to 2w, and then 7w again, infinitely loop). This can be avoided by untick the SpeedStep in the TS. But I am just curious what could be the potential cause of this problem? The throttle happens even below the temperature threshold.
Many thanks!

 

I have to say the new version is more powerful than ever, I am really enjoying being able to set static voltages for each profile as I test to get the best stability and temp with my 6700k. Job well done without a doubt!

If I could make a wish list for this beta I would put 2 features on it.
1. Set a startup profile just as you could do on the old version, this way if your voltage is too far out of wack you don't have to delete the ini and start over. Could be helpful with your everyday clock too.
2. Minimize back to system tray and not the system bar, this one is probably less important but just a little nice to have.

All in all thanks for the application, this version is a big leap forward.

 

@unityole - The first gen Core i did not include adjustable voltage. That is why ThrottleStop VID is grayed out and is probably why XTU does not work.

@GrofLuigi - Did you run the new version of ThrottleStop from a new folder or did you copy the new version into your old ThrottleStop folder? I am just trying to figure out if there is a problem with ThrottleStop or the new ThrottleStop is reading some previous settings from your old ThrottleStop.INI configuration file. I will put C States on the things to check out list. I was too excited about adding voltage control for Skylake so I have only been concentrating on that.

@xrender - There are multiple power limits. ThrottleStop is only accessing the one in the CPU MSR. There is another power limit that can be accessed via the Memory Mapped IO (MMIO). The WinRing0 driver that ThrottleStop uses removed that feature as far as I know so ThrottleStop cannot access this. XTU can access the MMIO register so that might be the difference. You could try using Read/Write Everything to access this register. I might have the address of this register written down somewhere. Most throttling schemes these days are power related and are not temperature related. No access to MMIO and no Core M hardware means that ThrottleStop is not going to be able to help you as much as I would like it to help you.

@GTVEVO - A default Start Up profile sounds like a good idea. I will see what I can do. Another idea when playing with voltages is to make sure ThrottleStop does not save anything to the .INI config file so if you crash with a BSOD, your unstable voltages will not be saved. The FIVR window already has this feature. For your second question, read the included ReadMe file. It tells you in there how to correct any icon issues.

That's what I like to hear.

@Mr. Fox - Overall, the bugs in Beta 6 related to Skylake are fairly minor. The voltage stuff should work 100% correctly so no worries there. Some of the bugs that I know about might have already been fixed but without hardware to test on, I am not yet sure. Hopefully I have some time and new hardware so I can keep working on this project. I will probably buy a 6600K but they are still in short supply where I live.

@Plur - Thanks for your offer to help. Skylake is a little different than Haswell so I need to sit down with some new hardware so I can better understand what the differences are. After listening to Intel, I thought FIVR and all the voltage controls were long gone but in reality, ThrottleStop works very similar on Haswell and Skylake and ThrottleStop accesses the voltages in the exact same way. I am looking forward to getting some new toys so I can do some hands on troubleshooting and keep this project moving forward. I don't really need a Skylake CPU but it seems like it is for a good cause. Don't tell my wife.

 

Thanks for the reply!
The one I know so far is MCHBAR+0x59A0 and the following 3 dwords (not sure if it is till valid for Broadwell). I am not sure if the way I calculate the address is correct. e.g. MCHBAR+0x59A0=0x48+0x59A0=0x59E8 so is it right that I just access 0x000059E8? I didn't see any changes in 0x000059E8 before and after modifying XTU though... Is there any source that I could know the offset for MMIO PL1 control?

 

The base address is typically 0xFED10000 and the long term PL1 Turbo Power Limit is at an offset of 0x59A0 so you need to add those two values together.

Here is an example of my Y510P.



It has 00000000 at FED159A0 which means this is not used to control this laptop. The Y510P uses the MSR power limit instead.

The one after that at FED159B0 shows

800002A8

The first 8 is the lock bit which means that this value cannot be changed. 0x2A8 = 680 decimal. This is the Current Limit and represents units of 1/8 Amps.

680 / 8 = 85 Amps

0x5990 is the maximum multiplier. 000000FF in my screen shot means that this location is not locked. It is set to the maximum value so it is not being used to limit my multiplier. If I set it to 00000018, my multiplier will drop down to 24 (0x18) and I will no longer be able to overclock.

Hope these "hints" help you solve your problem.

Edit - If your system is using the MMIO PL1, you might be able to set it appropriately and then set the lock bit to prevent it from being changed back.

 

It works So the XTU indeed modifies 0xFED159A0 to control the Core M TDP! Thx a lot
A side question: do you know if PECI throttling also programs this register or not? I once saw some EC src code sending PROCHOT to limit the processor frequency, but recently I also saw someone complaining that their computers PL1 are controlled by PECI. I am just curious if this is true. If so, is there a work around?

 

The lock bit is indeed a good idea. I will try to lock it and see if it will be further modified by some thermal management software.

 

Definitely try locking PL1 to the high power limit. That might be problem solved or that might force another throttling method to kick in.

PECI triggered throttling is another method but I have not looked into that one. Consumers are at the mercy of developers.

 

I always update the folder where TS is installed by binary compare. Usually only the EXE is changed. I keep a backup of the INI file and compare changes.

Definitely there is something weird going on with CStateLimit. Every time I enter the MAX dialog, the CStateLimit value reduces. Just now it reduced from 511 to 127. I exited TS and changed the value to 0xFFFF, then started TS and entered the dialog again and it reduced to 7777. Prior to this, I never edited the CStateLimit value, and, as I said, the PACKAGE CSTATE LIMIT button is grayed out all the time.

I edited it again to 0xFFFF, didn't enter the MAX dialog, and it remained at 0xFFFF, even after program exit.

 

Under Windows 8/8.1/10 this is always screwed up, but the notification area cleaner will straighten it out. I also create a Task to launch ThrottleStop after a 30 second delay because it will not start minimized to the system tray if it omit the 30 second delay with any of those craptastic operating systems.

I agree... best ThrottleStop release ever. Really sweet.

There, I fixed that for you, bro. Developers that do awesome things are talented professionals, such as yourself. Unfortunately, I cannot say anything kind about the rest of them.

Still looking really great. I helped another member of the forum get ThrottleStop set up with profiles and hotkeys, etc. today. Everything works very nicely with Skylake and I couldn't be happier with how this new version plays so nicely with 6700K.

 

@unclewebb As Usual, awesome work. Within 1-2 months I probably have an new laptop(Clevo P870 )with Skylake. Great that you've added support for Skylake in Throttlestop. It will be a pleasure to have your ingenious software on my next laptop as well. Continue with the job you are doing. The job you do is appreciated by numerous people who value performance. Laptop OEM's haven't won yet. Thanks

 

Didn't read the rest of the post, or the original post, however I CAN confirm launching TS8 beta 6 for the first time today (was using TS8 beta 3 from before; I didn't even touch 4 and 5 because lazy) and as soon as I launched TS8 b6, it had my previous default profile set (the FIVR settings) and the TPL settings saved. It did change "task bar" and "bd prochot" being ticked and my "options" setup etc. The other three profiles are matched to my previous "default" profile. My TSBench records were wiped as well.

So if he is having any issues with previous data saved and being used, it's definitely possible that it's using it. He might want to set back his date, launch his previous throttlestop, reset whatever he wants to default, then launch beta 6 again after. Or, he can wipe the settings pages for beta 6 and let it start fresh.

 

@GrofLuigi - If the Package C State Limit button is grayed out, that means the bios has locked this register so it cannot be adjusted by ThrottleStop or any other software for that matter.

When you run ThrottleStop and there is no ThrottleStop.INI configuration file or if ThrottleStop does not find the CStateLimit line in the INI file, it should check what your CPU is currently set to and then it should use that value for all 4 profiles. If an older version of ThrottleStop put an incorrect value in the INI file or if you are manually editing the INI file, this might be causing problems. 0xFFFF is not a valid value for your CPU. If you run ThrottleStop and you do not enter the C State window then ThrottleStop is not going to change that value in the INI file. It will just leave it as is at 0xFFFF. If you open up the C State window and click on OK, ThrottleStop is going to save a new value to the INI file. If 0xFFFF is not a legal value then ThrottleStop changing this to 0x7777 sounds reasonable.

Run Notepad with Admin rights so you can edit the INI file. Exit ThrottleStop, open up Notepad and delete the CStateLimit line. Save the changes to ThrottleStop.INI and then start up ThrottleStop. Open up the C State window and click on OK. Have a look in the INI file and see what value was saved for the CStateLimit. Open and close ThrottleStop a few times and go into the C State window and click on OK a few times and change profiles back and forth and see if this value changes to something unusual. The bios has locked the C State Limit on your CPU so this value should stay the same. Do not manually edit the INI file with random values.

A value of 7 is the code that means your CPU has no Package C State Limit. ThrottleStop calls that MAX. A value in the INI file of 7777 means all 4 profiles are set to the same MAX value. Manually setting this to FFFF is not recommended.

I will do some more testing of this on Tuesday.

 

Thank you for this very useful software. Could you please link the Turbo Power Limits with profiles? They seem global right now.

I have a Dell Venue 11 7130 tablet (i5-4210y) and I'd like to be able to create a profile that limits the Package Power to maybe 4W or so (like the Atom Z series chips), for extended battery life when performance is not critical.

 

Deleted the CStateLimit line (with TS not running). Upon entering TS's dialog/window it said c6r on the button (which was previously MAX - both the button to enter it (between BCLK and DTS buttons) and the button inside the window (below Package C State Limit line). That's why I always called it MAX. After exiting TS, the line reappeared and reads CStateLimit=0x3333. But I didn't reboot after stopping TS the first time, so maybe it had previously read/set the flag incorrectly?

A long time ago I tried recreating the INI file from scratch (deleting the current INI file), but I couldn't recreate the settings exactly as they were, or as I wanted them, so I gave up and from then on I always reuse the INI file.