Anyone tried throttlestop with R2's yet?

I tried it w/ M1730 and got 200mhz in boost, but I'm still not too familiar w/ the software.

Has any one tried it yet with R2 yet? Any magic tricks it can do like it does with X9000 cpus?

Thanks.

 

I don't believe it supports Core series processors and I don't think a similar program exist for the Core series yet.

Core 2 Duo IDA is similar to the Core Series Turbo Boost, but the Core Series isn't limited to one CPU. X9000's have unlocked multipliers and the regular Core Cpu's do not.

 

Ah I see. I thought it be possible since we are able to OC(if you call it that...).

My T7700(I think) w/ locked multipliers was still able to gain 200mhz, too so was wondering if anything was possible w/ it.

Thank you sir.

 

The M11x can overclock because there is a setting in the bios to adjust the FSB.

Your T7700 isn't really overclocked. Your T7700 has IDA (Intel Dynamic Acceleration). If one core is idle the other core can accelerate and extra 200MHZ. Throttlestop has a mod that allows for dual IDA, so instead of only one core accelerating, both cores do. The chip frequency isn't really going any faster than it was designed/set to, but its nice because both cores can now accelerate 200mhz or be fixed at 200mhz faster.

 

Old thread but just wanted to add that ThrottleStop works with the newer Core i CPUs. If you right click on ThrottleStop there are options to adjust the TDP/TDC values which can improve performance by adjusting the amount of turbo boost you get. The M17x users are having lots of fun with this new feature when used with their 920XM and 940XM CPUs.

 

Thanks, UW. I gave it a shot a couple of weeks ago and couldn't at that time tell that it was doing anything. I'll revisit it again.

 

when i set the maximum TDP to 11W (wont go higher) and then the amps to 40 and press OK it will reset back to default 10W/30A. I have the i5.

disable turbo boost doesnt seem to work either.

 

Doesn't seem to work for me either. I tired ThrottleStop 2.76, starting to think m11xr2s have a throttling issue similar to the m15x..

 

I think you may need a Xtreme processor to be able to change those settings. Are there any reports of regular Core series processors being able to change these settings.

 

I'll definitely keep an eye on this thread. Hope people get it working on the R2s...

 

bump... anyone got this working... this could be huge for us! maybe we can lock it at max turbo frequency and get even MORE CPU POWA.

 

bump any luck?

 

Another bump, I still have throttling issues.

 

What throttling issues are you having? The only one I know of is related to the GPU not the cpu. You can fix that one by disabling nvidia powermiser.

 

I guess no one seems to realize, but try to have HWinfo or CPU-Z running on an external monitor and watch the clock speeds. They constantly jump up and down, sometimes down to the rare 6x ratio, but mostly from 9x-14x. I've posted throughout the forum numerous times, and it's been acknowledged but not much has been done about it. I've tried to fix it, but haven't had much luck; hopefully someone else can though.

The cpu throttling issues have existed on other Dell laptops and was a big issue on the m15x.

 

What? That isn't an issue, the changes in clock speeds/multipliers are a feature to save battery life!
SpeedStep - Wikipedia, the free encyclopedia

 

Perhaps you misread. When playings games, the cpu throttles down from 16x down to anywhere from 9x-14x. It rapidly varies, as in .5 seconds it's at 9x, 1 second 12x, 1.5 seconds 10x, 2 seconds 14x, and so on.

In general Windows usage, the multipliers vary from 14x-the rare 17x, but they're mostly at 16x. If I watch closely though, they sometimes drop to 5x for a split second; this might or might not be a glitch in HWinfo.

In short, during gaming multipliers are never constantly locked, thus you see a variation in FPS.

 

Are you watching your resource monitor? The multiplier should only reduce while the resource usage is low or erratic. I'm going to go ahead and run some tests on mine

 

You're only looking at half of it, and the half that's irrelevant to the thread . Speed Step is made to lower the multiplier, as you say, but you can disable it in the bios to prevent it from decreasing lower than the default multiplier. Thus, it is not the issue of the thread--because anyone who wants it off can do so already. The issue of the thread is Turbo Boost. Turbo Boost is nearly Speed Step in reverse; that is, it increases the multiplier higher than the default multiplier. The point of this thread is to put the multiplier up as high as Turbo Boost allows and lock it there. Throttlestop is meant to lock the multiplier where you want, but it's currently not doing that for the M11x.

There is more to it than you say. There are timings that can cause lag between the need for more speed and the actual increase of the multiplier. One is the timing for how often the load on the CPU is checked to see if it should increase or decrease, and if it's set too low, then it won't change as often as needed. There is another determining how long the load must be at a lower amount before the CPU decreases the multiplier. If it's too low, in conjunction with the first check mentioned, it can cause a bit of lag. Unfortunately, there is no way to change these natively in Windows (there is in Linux, at least for Speed Step--I'm not sure if there is for Turbo Boost). As a result, it would be very handy when running games or other intense software to be able to lock the CPU at the highest multiplier.

 

corwinicre, I applaud you. Thank you for explaining the problem way better than I can, spot on

 

9x133 = 1200mhz. The i3/i5/i7 UM cpu has a base clock of 1200mhz. When an application puts a full load on the cpu it shouldn't be lower than a 9x multiplier. Turbo boost will increase the clock of the cpu based on the number of threads, load, and temperature. The cpu multipler should be changing between 9 and 16x. This is not a throttling issue, but a design feature called Turbo boost.

Some other Dell had throttling issues caused by an insufficient power adapter. The M11x R2 comes with a 90w which should be enough for a ULV cpu and a midrange cpu like the GT335m.

When I turn logging on in core temp and play TF2 I typically see the cpu in the 1600-1800mhz ran, its usually pretty constand at 1733.

 

You're missing the point, we want Throttlestop to work for the m11x so that we can keep a constant locked multiplier of at least 16x, making performance smoother. Turbo boost shouldn't be kicking back and forth wildly between frequencies, or it would be a lot better if it didn't.

 

Yours was perfectly fine, too. I've wasted so much time messing with the configuration for cpufreqd, a Linux daemon to control cpu scaling, because my previous laptop was constantly overheating that I take every opportunity I can to talk about the stuff so that all that time spent on it feels like less of a waste . (Needless to say, I love the lower temps of the M11x.)

Everything you say is true, but there is no reason to not be able to lock the multiplier higher, even by your own facts there. As you say, your cpu runs pretty constantly at a high multiplier and has no trouble. In my experience, I have never had the fan max out its speed like it does when flashing the BIOS. This shows the system can handle the heat put out with the CPU "boosted". I understand it is not possible to lock the first core at its highest and the other at its highest (since the first core reaches a higher speed by taking the throughput that the other is not using), but one should be able to lock all of them at the second core's highest, at least.

At a minimum, I'd like the ability for it to stay at the highest frequency the temperature and core use allows (i.e., do not drop the multiplier just because there isn't a high load on the cpu). That would prevent some slight fps drops in games, as miXwui said above. In defense of not being able to lock the multiplier, you argue that TB already effectively manages it. Let us look at that assumption. How good is TB really at matching frequency to performance? Check out this paper: http://www.cs.sfu.ca/~fedorova/papers/TurboBoostEvaluation.pdf. (Note: This paper uses Linux, not Windows, as the test system, so it is potentially not exactly analogous to how it works on Windows, but I cite it for the potential of TB to not be used correctly.) Skip to Figure 9 on page nine for the cliffnotes version. You can see by how often TB drastically differs from the actual load ("utilization") of the core.

It would be even better to be able to control the number of cores in use and then put it on a per-core basis at the highest available frequency--that'd be great for games that aren't designed to use multiple cpus/cores. Going to page eight of that article, talking about Figure 9 again, the authors say,

And continuing later,

If one could make additional cores inactive--or, at least, more often inactive--then it would "boost" the first core as high as possible more often. The argument you give that it is designed to increase the frequency when needed and decrease it when not needed is potentially invalidated not because it doesn't do as you say but because it doesn't do it well . If the user could lock frequencies of the different cores, it could help (it'd be even better to be able to change the thresholds to cause all but the first core to scale down more often so that they wouldn't have to be locked at one frequency, which also would improve TB for programs that use a single core often, but let's take it one step at a time ).

Now, there's one thing you might argue against me there. The reason the first core can scale up as high as possible is because the other cores can scale down, yet they can't scale down if the multiplier is locked. Yes, this means in some cases it is better to have Turbo Boost free to do its thing rather than have the multipliers locked, but it is not always better. With programs that take advantage of multiple cores, it is probably rare that it would be better for one core to have that extra 133 mhz over having all the cores at the frequency they are all capable of (i.e., the speed 133mhz less than the first core can do). With the ability to lock the frequency, the user can lock the multiplier when it is best to do so and leave TB to manage it when it's not. Choice is always better.

Along with the ability to lock the multiplier comes not only keeping the cpu faster but also slower. In other words, locking it disables Turbo Boost. (Yes, you can do it in the BIOS, but being able to do it without rebooting every time you switch between battery and AC would be much nicer.) Why is that a good thing? The inability of Turbo Boost to effectively manage power is mentioned in the given article often, each time for a different reason. For one, they deny the claim some people have made on the forum that Turbo Boost causes execution time to be lower, thereby providing better battery life:

They go into more detail later:

In other words, although Turbo Boost is praised for doing things on a per-core basis, it does not always do it well, both by not "boosting" the active core enough nor by lowering the inactive cores. Considering that the best way to act is going to change with any program running and with any set of programs, the idea that TB as used by Windows (or any OS) is always going to be the best way is irrational.

tl;dr: Power--err, choice--to the people! <raises closed fist>

 

I'm not saying that turbo boost = maximum performance. I'm saying that there is not a throttling problem/issue as miXwui says there is. This behavior is not a Dell problem or a problem at all, this behavior is as designed by Intel. Its called turbo boost!

Yes you can get more performance by overclocking or locking the cpu speed at a high clock (if the app worked for the UM processors), but that doesn't make the cpu's current behavior a thottling issue as miXwui said. My point is simply this behavior is not the same as the throttling issues other Dell's like the XPS 16 have experience because of bios issues and a lack of power for the AC adapter.

 

^Whoops, my mistake. I understand now. Sorry about that.

 

I've never liked Turbo Boost. The CPU itself checks a number of thermal conditions that need to be met in order for it to increase the frequency. It means the frequency is likely never to be stable for a longer time. It means fluctuating performance in applications and games.

People should, in my opinion, look at the base frequency of the CPU. That's your safe performance. The maximum Turbo Boost frequency is irrelevant to me.

 

MaxGeek, by throttling I meant Turbo Boost does dumb things, Turbo Boost IS essentially just automatic throttling of the cpu. So, technically there IS a "throttling issue." It's just a matter of semantics

But that's besides the point, what we all would like is to be able to lock the cpu in the ways that corwinicre pointed out. Thus, either performance, energy saving, or a balance between the two could be locked. And this IS possible, Throttlestop works on i7s, just not the m11x yet

Mackan, if everyone were to disregard Turbo Boost and only look at the stock frequency, that would be 1.2ghz. That's lower than the r1's default! The selling point of the icores is Turbo Boost, so I think the frequency is pretty relevant; the ghz at 16x is what I would like to be "safe performance."

 

There seem to be some misconceptions about these CPUs. If your CPU is not an Extreme CPU, that means the maximum multiplier is locked and you can not use any software to make it do something that Intel did not design it to do.

A Core i7-640UM has a base multiplier of 9. To get maximum turbo boost from a Core i CPU, you need to set the multiplier request register to 1 more than the default multiplier. For this CPU that would be a ThrottleStop setting of 10. Software doesn't have any control over the amount of turbo boost the CPU will give you. That's up to the CPU based on how many cores are in the active state and whether the CPU is operating under the TDP/TDC turbo power limits. The Extreme CPUs let you increase these limits. Some other Core i dual core CPUs like the i7-620M let you increase these limits but that depends on whether the motherboard supports that feature. If this has been disabled at the bios level then there is nothing software can do.

The i7-640UM has a maximum multiplier of 17 but that only happens when one core is in the active state. As soon as a second core becomes active to process a background task, the maximum multiplier for both cores will drop down to 14. If the CPU is operating above the turbo TDP limit then you might not get any turbo boost and will be limited to the default multiplier which is only 9. These are limitations of this CPU, not limitations of ThrottleStop.

Turbo boost is very dynamic and the multiplier can be changing back and forth like this hundreds of times a second. Some monitoring software is misleading when this is going on. ThrottleStop shows the average multiplier for each thread and should be extremely accurate reporting these CPUs because it is one of the few programs that follows the method outlined by Intel to accurately monitor these CPUs.

If you try to adjust TDP or TDC but it is blocked or immediately gets reset to default values then I guess this feature is not supported by the current bios.

Can someone post some screen shots of ThrottleStop? How about download wPrime and run a benchmark and set the number of threads to 1 then 2 then 3 then 4 and take a screen shot of each example about half way through a 32M test. That will let me see if ThrottleStop is working correctly and correctly reporting these CPUs. So far all I've read is the desire for these CPUs to run at a performance level that they are physically not designed to run at constantly. It would be nice to lock the multiplier at a constant 14 or 17 on both cores but that's a physical impossibility for the i7-640UM.