How to Supercharge the M11x with Core i5/i7 UM CPUs

The Alienware M11x R2 uses Intel's Ultra Low Voltage Core i5 and Core i7 UM processors. These CPUs are designed to limit the amount of turbo boost when fully loaded which reduces power consumption but that can also significantly decrease performance.

There are two values in the CPU called TDP and TDC that controls when the CPU is allowed to use the turbo boost feature. If your CPU is operating below the turbo TDP power limit and is also below the turbo TDC current limit, you get full turbo boost. If you are over those limits, turbo boost will not be used. The CPU is constantly checking power consumption and adjusting turbo boost accordingly.

If you would like to increase the amount of turbo boost at full load, you need to increase the turbo TDP limit that the CPU is using.

WARNING: Increasing the TDP limit can increase power consumption and can reduce battery life when running on battery power. Any damage caused by using software like ThrottleStop to change this limit is at your own risk.



Download ThrottleStop from the link in my signature.
Click on the TPL button to open up the Turbo Power Limits window.

Most UM processors have TDP set to 10 watts and TDC set to 30 amps.



To get more full load turbo boost you need to increase TDP to 30 or higher.



When finished, click on OK to save your settings.

You can also set ThrottleStop up so it uses different TDP/TDC turbo boost limits for each profile. This will allow you to use your 30/30 limits when gaming and you can use the default 10/30 limits when you are on battery power so there will be no change in your battery life. Use the Options window to set a default AC profile as well as a default Battery profile so this will change automatically.

Increasing the turbo TDP limit within the CPU will help to maintain full turbo boost even when fully loaded. In these UM CPUs, this can increase full load performance by up to 50%. Here's an example of the increase in CPU performance that is possible.



There is also a setting in the Options window called Force TDP/TDC. This controls how often the TDP/TDC settings will be checked and adjusted in the CPU. Setting this value to a smaller number increases the frequency of this. Going too low can reduce performance. The default setting of 16 gives good performance without slowing down the CPU. User testing has shown that a value of 8 or less might help you to maximize your overall CPU performance. If you want to adjust this to maximize performance then do some testing. Use a program like wPrime and you can adjust this value based on your benchmark scores from this program. The quicker wPrime can do a 32M or 1024M calculation, the faster your CPU is running.

The Core i7-920XM and Core i7-940XM in the Alienware M15x and M17x do not need the Force TDP/TDC option set to a low number. Setting this to 1000 or higher should be OK for them to maximize performance. So far, this has only been tested on the M11x but there might be other laptops with Ultra Low Voltage UM processors that can also benefit from this setting.

If your CPU fan is clean and your heatsink is properly attached, the M11x will not use Clock Modulation throttling or Chipset Clock Mod throttling so there is no need to check either of these boxes.

The Set Multiplier box seems to be optional for the M11x. If you use this feature, you need to set it as high as it can go. Keep increasing this until it says Turbo.

The numbers in the FID column represent the average multiplier on each thread in real time. When running a wPrime benchmark and all threads are fully loaded, the average multiplier value in this column should increase after TDP has been adjusted higher. This indicates your CPU is running faster.

Increasing TDP/TDC will show the biggest gains if you are overclocking your CPU in the bios by raising the BCLK value. Using ThrottleStop to increase the TDP/TDC limits may show little to no performance gain if you are not overclocking in the bios.

It might also be possible to decrease the TDP/TDC settings if you are interested in reducing power consumption and extending battery life. This possible use of this new ThrottleStop feature has not yet been investigated.

WiFi Disconnect Problem

When using this new ThrottleStop feature, mcham was having a problem with his Dell 1520 WiFi mini card losing its connection after about half an hour of gaming. His laptop stayed connected when on a wired connection and it also seems to be OK since he switched to an Intel 6300 Ultimate N WiFi mini card. I'll update this info if he is still having problems. Update: this may just be one bad card and not a common problem with the Dell 1520. Other users have had no problem gaming for hours with the Dell WiFi card while using ThrottleStop.


Start ThrottleStop with Windows
http://forum.notebookreview.com/6865107-post2.html


TS Bench Results
MasivB - Core i7-640UM - 167.0 MHz - 38.457 seconds

wPrime Results
-

++++++++++++++++++++++++++++++
NOTE FROM AW FORUM MOD:

While some of you will want to explore what Webb has outlined above, please note
that doing so may result in decreased battery time. Also, and this is just a guess
after dealing with tech support over the years, support will not work with you in the
event of any errors, BSODs, or general troubleshooting. The reason for this is simple -
using TS modifies the performance of the system (BIOS) from the way it was originally
designed.

This thread is a 'modification' thread. Be sure you note the warning that Webb adds
to his app, 'use at your own risk'.

Webb cranks out some very useful stuff and I thank him (as we all do) for his continued work on TS.

For those of you who are new to all of this, ask questions BEFORE jumping in. If you are happy with
your 11x the way it is, then don't worry about it. Bottom line, know what you are doing before you do it.

Comments on this message need to be PM'd directly to me. Thanks.
-BB

++++++++++++++++++++++++++++++

 

i love you

 

I look forward to what you find on this. If i had more time i would help you out. mabey on the weekend if other plans fall through.

 

Thanks, unclewebb! I'll download and test later this evening.

 

Glad to see you over here Uncle. Does it need to be WPrime or can I use LinX. Is there a way to disable HT via code? I wouldn't feel so bad about maxing my TB given the lmited cooling if HT were also disabled. Nonetheless, I'm game for testing it.

Oh yeah, RealTemp works great on our M11x's if you guys haven't tried it.

 

I did as you said with my Core I5 520UM, revision C2, no overclocked fsb, and the TDP and TDC settings were not locked in Throttlestop.

First time I ran through 32m tests, I forgot that the cpuz packaged with wPrime is messed up and doesn't unload, taking up about 25% of the cpu the entire time, so the times were messed up. For anyone who tries this, remember to go into Task Manager and kill cpuz, or else your times will be off.

Here are my results after killing cpuz:

tdp tdc time_to_run_32m
(control with "balanced" power profile)
10 30 36.116
(control with "performance" power profile)
10 30 36.067
(with ThrottleStop)
20 60 37.043
40 120 36.408
80 240 36.436

As you can see, there is no real difference, despite increasing the TDP and TDC values 8-fold.

Have you seen TuneUp Utilities? In this thread, revdiesel posted how the program TuneUp Utilities has an optioin to keep Turbo Boost on full, except it drops during games and such--perhaps because it doesn't change the TDP and TDC values, in which case your theory would seem to be right. I suspect, then, the reason there is no change in the 32m test is Turbo Boost kicks on the entire duration of the 32m wPrime test. If that's the case, there will only be a noticeable difference when Turbo Boost is needed for a longer time, like gaming or while running a longer benchmark.

 

The original testing was on an Alienware M11x with a Core i5-520UM. wPrime times improved by 35%.

Without any hardware in my hands, that was just my first stab in the dark with this. Today's algorithm is slightly different. I think it is possible to get the performance increase when fully loaded up over 40% for some of these UM CPUs. Lots of potential just waiting to be tapped. I'm excited and I don't even own one of these.

I believe HT can only be turned off at the bios level so there isn't any software that is going to be able to turn that off for you.

You can use whatever you like for testing purposes. I recommended wPrime because it is multi-threaded and any performance increase will immediately translate into the same percent decrease in wPrime times. That benchmark if 100% CPU dependent.

corwinicre: Was ThrottleStop enabled when you were testing?

 

Yes, ThrottleStop was enabled for all the tests after where it says "(with Throttlestop)". In fact, to be sure that the TDP and TDC values were being applied, I disabled and re-enabled it each time I changed the values.

Since Turbo Boost stays on the entire 32m wPrime test without ThrottleStop running, how could be any speed improvement shown on the 32m test with it?

 

The 520UM tested yesterday was not giving any turbo boost when fully loaded while running wPrime. What multiplier does ThrottleStop show when running wPrime with TDP/TDC set to the default values?

Can you post a screen shot of CPU-Z and ThrottleStop while wPrime is running?

 

I have the i5 as well as i've noticed a 130 point increase in my 3dmark06 CPU score!!!!

wprime does not work for me for some reason never gets past checking the hardware using CPUZ.

i set it to TDP/TDC to 30/70

The left picture is without throttle stop the right is with 30/70 enabled AMAZING WORK unclewebb!!

 

I had issues with wPrime too but when I moved it into the Program Files folder then it worked fine.

I'm glad to see you are seeing some improvement. You could also try bumping up TDP to around 50 and see how that works out.

 

Something is wrong with your test system, then, because TB has always kicked on for wPrime benchmarks (and other benchmarks) for me, and my scores in the benchmark thread here aren't any better than other people with the I5. If TB didn't kick on for a <50 second test, it would be quite useless. Perhaps it was limited via the power plan that was chosen because it can be turned off in the power settings, and then ThrottleStop was overriding the power plan.

Here's a pic like you ask to prove it (linked due to its size) http://imgur.com/HexNe.png

I just watched cpuz without Throttlestop running, and it showed TB active during the wPrime 32m test.

Edit: glad to see it's helping with longer benchmarks!

 

Unclewebb, I'm going to run the tests now for you. But would you prefer O/C FSB on or off? I am very lucky in that my m11x i7 O/C's stable to 166Mhz

 

M11Ash: Use whatever settings you like. The more different data the better.

corwinicre: It was not my test system. I was helping an inexperienced user with his M11x yesterday and he was getting zero turbo boost when his i5-520UM CPU was fully loaded. His CPU was running at the default 8 multiplier and that was it. When a CPU is fully loaded, the power profiles shouldn't make any difference as the CPU should be running as fast as possible.

Can you use ThrottleStop to return your CPU to default TDP/TDC settings and show a screenshot of ThrottleStop while wPrime is running? I don't understand why your CPU gives you full turbo boost while his CPU was giving him none. A different stepping maybe?

Here is part of his log file:

Code:

  DATE     TIME    MULTI   C0%   CKMOD  CHIPM  DTS   GPU
10/10/10 00:06:18  10.27    7.6  100.0  100.0   50    52
10/10/10 00:06:18  11.52   13.9  100.0  100.0   50    52
10/10/10 00:06:18   7.97    8.0  100.0  100.0   50    52
10/10/10 00:06:18  10.19    7.5  100.0  100.0   52    52
10/10/10 00:06:18   9.61    6.2  100.0  100.0   50    52
10/10/10 00:06:19   9.73    7.0  100.0  100.0   51    52
10/10/10 00:06:19   8.87   75.4  100.0  100.0   42    52
10/10/10 00:06:19  10.94  100.0  100.0  100.0   44    52
10/10/10 00:06:19   8.00  100.0  100.0  100.0   45    52
10/10/10 00:06:19   8.00  100.0  100.0  100.0   45    52
10/10/10 00:06:19   8.00  100.0  100.0  100.0   45    52

As soon as the C0% load went up to 100.0%, his multiplier dropped down to the default 8.0 multiplier and was stuck there.

Is there an economy power profile that he might have been using that disables the full load turbo boost?

That's the problem with the Core i7-720QM. When all 8 threads are fully loaded, it gives you zero turbo boost because it is operating beyond the turbo TDP/TDC limits. You're right, that is useless and those CPUs have the TDP/TDC settings locked so there is no way to fix that problem.

 

Here is a picture of my results. CPU is i7 (Stepping 5, Revision K0), overclocked to 164MHz FSB, High Performance setting, on battery. TDP and TDC set to double the old value.

UPDATE: Also tried tripling the TDP and TDC values as well, but performance increase is negligible compared to doubling the values.

 

Thanks zephir. Is that a 620UM? What TDP/TDC values did you use?

 

It's 640UM. TDP is set at 20W, TDC is at 60A

 

i5 @150

1st run TS off 4 threads
2nd run TS on 4 threads double TDP/TDC to 20/60 from stock 10/30
3rd run TS on 4 threads TDP/TDC 30/90
4th run TS off 2 threads
5th run TS on 2 threads TDP/TDC 20/60
6th run TS on 2 threads TDP/TDC 30/90

 

OK some help please guys, wPrime can't seem to find my hardware its just stuck on a loading loop when I start it up.........

 

What stepping and revision of the i5 do you have?

 

I just ran this on my system.

Core i7UM @ 162
Ran 4 consecutive 32MB wPrime benches. All results were very consistent at 35 seconds.
Fired up TS, doubled the TDP and TDC values (20/60 respectively)

This dropped my wPrime time down to 27.6 seconds. That's pretty significant in my book. Don't have a lot of time at the moment to test further, but here's the screen shot where doubling the values dropped my score by 21%.



Here's a dump of the log file I just captured with another run (this one was 26 seconds and is not pictured above)

Code:

  DATE     TIME    MULTI   C0%   CKMOD  CHIPM  DTS
10/11/10 17:24:19  10.45    6.8  100.0  100.0   56
10/11/10 17:24:20  12.00    8.2  100.0  100.0   54
10/11/10 17:24:21  11.10    7.1  100.0  100.0   54
10/11/10 17:24:22  10.52    7.2  100.0  100.0   55
10/11/10 17:24:23  10.76    6.6  100.0  100.0   54
10/11/10 17:24:24  10.15    5.9  100.0  100.0   54
10/11/10 17:24:25  11.17    7.9  100.0  100.0   55
10/11/10 17:24:26  11.82    8.0  100.0  100.0   54
10/11/10 17:24:27  10.74    7.0  100.0  100.0   55
10/11/10 17:24:28  11.98    7.7  100.0  100.0   54
10/11/10 17:24:29  12.32    8.7  100.0  100.0   54
10/11/10 17:24:30  11.60    8.0  100.0  100.0   54
10/11/10 17:24:31  12.79    8.9  100.0  100.0   54
10/11/10 17:24:32  12.73   10.4  100.0  100.0   55
10/11/10 17:24:34  12.20    8.4  100.0  100.0   54
10/11/10 17:24:35  12.56   11.4  100.0  100.0   54
10/11/10 17:24:36  11.39    7.3  100.0  100.0   55
10/11/10 17:24:37  11.96    8.7  100.0  100.0   55
10/11/10 17:24:38  11.15    7.2  100.0  100.0   55
10/11/10 17:24:39  11.23    8.7  100.0  100.0   55
10/11/10 17:24:40  11.60    7.1  100.0  100.0   54
10/11/10 17:24:41  12.11    8.1  100.0  100.0   55
10/11/10 17:24:42  10.98    6.5  100.0  100.0   56
10/11/10 17:24:43   9.25    4.2  100.0  100.0   56
10/11/10 17:24:44  10.74    4.9  100.0  100.0   56
10/11/10 17:24:45  10.93    4.9  100.0  100.0   56
10/11/10 17:24:46   9.93    4.3  100.0  100.0   56
10/11/10 17:24:47  11.55    5.3  100.0  100.0   57
10/11/10 17:24:48   8.87    4.4  100.0  100.0   55
10/11/10 17:24:49  10.27    5.2  100.0  100.0   55
10/11/10 17:24:50  11.41    4.6  100.0  100.0   56
10/11/10 17:24:51   9.25    4.3  100.0  100.0   56
10/11/10 17:24:52  10.44    6.3  100.0  100.0   55
10/11/10 17:24:53   9.69    4.8  100.0  100.0   56
10/11/10 17:24:54  10.84    4.9  100.0  100.0   55
10/11/10 17:24:55  10.46    4.9  100.0  100.0   55
10/11/10 17:24:56   8.70    4.4  100.0  100.0   56
10/11/10 17:24:57  11.15    5.0  100.0  100.0   56
10/11/10 17:24:58   8.57    4.1  100.0  100.0   56
10/11/10 17:24:59  10.50    4.9  100.0  100.0   56
10/11/10 17:25:00   9.94    5.1  100.0  100.0   55
10/11/10 17:25:01  10.55    5.9  100.0  100.0   55
10/11/10 17:25:02  13.44   18.2  100.0  100.0   54
10/11/10 17:25:03   9.21    4.7  100.0  100.0   57
10/11/10 17:25:04  10.56    8.5  100.0  100.0   51
10/11/10 17:25:05  13.45   13.8  100.0  100.0   55
10/11/10 17:25:06  11.13    5.3  100.0  100.0   55
10/11/10 17:25:07  10.18    5.5  100.0  100.0   56
10/11/10 17:25:08   9.03    4.7  100.0  100.0   56
10/11/10 17:25:09  10.77    5.2  100.0  100.0   56
10/11/10 17:25:10   9.96    4.6  100.0  100.0   56
10/11/10 17:25:11  10.77    4.9  100.0  100.0   55
10/11/10 17:25:12  10.39    5.1  100.0  100.0   54
10/11/10 17:25:13   8.85    4.2  100.0  100.0   55
10/11/10 17:25:14  10.74    4.7  100.0  100.0   56
10/11/10 17:25:15   9.87    4.4  100.0  100.0   55
10/11/10 17:25:16  11.57    4.9  100.0  100.0   55
10/11/10 17:25:17  11.58    5.2  100.0  100.0   55
10/11/10 17:25:18   8.54    3.8  100.0  100.0   56
10/11/10 17:25:19  11.31    4.9  100.0  100.0   56
10/11/10 17:25:20   9.15    4.0  100.0  100.0   56
10/11/10 17:25:21  11.16    4.7  100.0  100.0   56
10/11/10 17:25:22  10.98    5.7  100.0  100.0   55
10/11/10 17:25:23   8.94    4.9  100.0  100.0   55
10/11/10 17:25:24  11.23    4.9  100.0  100.0   56
10/11/10 17:25:25  10.21    4.6  100.0  100.0   56
10/11/10 17:25:26  11.19    4.8  100.0  100.0   55
10/11/10 17:25:27  11.16    5.3  100.0  100.0   55
10/11/10 17:25:28   8.89    4.1  100.0  100.0   56
10/11/10 17:25:29  11.01    4.7  100.0  100.0   56
10/11/10 17:25:30   9.65    4.3  100.0  100.0   55
10/11/10 17:25:31  10.00    4.6  100.0  100.0   56
10/11/10 17:25:32  10.09    5.4  100.0  100.0   55
10/11/10 17:25:33   8.78    4.4  100.0  100.0   56
10/11/10 17:25:34  10.16    4.7  100.0  100.0   56
10/11/10 17:25:35   9.09    4.5  100.0  100.0   57
10/11/10 17:25:36  10.90    5.0  100.0  100.0   56
10/11/10 17:25:37  11.45    7.5  100.0  100.0   55
10/11/10 17:25:38   9.94    6.2  100.0  100.0   55
10/11/10 17:25:39  11.67    6.8  100.0  100.0   56
10/11/10 17:25:40  10.86    5.6  100.0  100.0   56
10/11/10 17:25:41  13.29   24.1  100.0  100.0   43
10/11/10 17:25:42  13.70  100.0  100.0  100.0   38
10/11/10 17:25:43  13.83  100.0  100.0  100.0   37
10/11/10 17:25:45  13.58  100.0  100.0  100.0   41
10/11/10 17:25:46  12.64   99.9  100.0  100.0   37
10/11/10 17:25:47  13.51  100.0  100.0  100.0   37
10/11/10 17:25:48  13.68  100.0  100.0  100.0   36
10/11/10 17:25:49  13.51  100.0  100.0  100.0   36
10/11/10 17:25:50  13.56  100.0  100.0  100.0   35
10/11/10 17:25:51  13.63  100.0  100.0  100.0   35
10/11/10 17:25:52  13.29  100.0  100.0  100.0   35
10/11/10 17:25:53  13.49  100.0  100.0  100.0   36
10/11/10 17:25:54  13.53  100.0  100.0  100.0   34
10/11/10 17:25:55  13.48  100.0  100.0  100.0   34
10/11/10 17:25:56  13.15  100.0  100.0  100.0   35
10/11/10 17:25:57  13.56  100.0  100.0  100.0   35
10/11/10 17:25:58  12.73   99.9  100.0  100.0   35
10/11/10 17:25:59  13.59  100.0  100.0  100.0   33
10/11/10 17:26:00  13.56  100.0  100.0  100.0   35

 

That's never worked with wPrime for the R2. Earlier versions had a hidden cpu-z process that would get stuck and take up 50% (or was it 25%?) of your CPU cycles in the background. With this new version it still doesn't detect R2 hardware, but at least they fixed the rogue cpu-z process that was killing people's scores with wPrime.

Just hold the Shift key down to bypass the hardware detection.

 

Mine does the same thing..........hold down SHIFT and it will bypass hardware

 

Yep, if you set the Maximum Processor State low in the Advanced Power Settings, it will keep it underclocked or simply with Turbo Boost off. For instance, with the Maximum and Minimum Processor States at 5% in my Power Saver plan, notice how my multiplier doesn't go above 5x even while running the 32m test: http://imgur.com/WIZf0.png. That's why I guessed he was running a power saving profile. I've changed mine so many times I can't say if the default one would keep it at 8x.

My stepping listed in CPU-z is 2 (this is beyond my understanding, so bear with me if you mean something else).

Here's the pic you ask for (you meant with Throttlestop off but with the screen open, yes?): http://imgur.com/tSXLt.png. I noticed it dropped from 13x to a constant 12x around the time everything in the C0% column reached 100%, which was a second or so after starting the 32m test.

 

The one UncleWebb linked to still has the cpu-z process that stays loaded taking up 25% of the cpu--or at least the one I downloaded from the page, version 204, does. If anyone else is using that, bring up Task Manager and kill cpuz153.exe.

 

Guess that's because I was running CPU-Z (155) already. So, yeah. Everyone needs to kill that stupid CPU-Z process each time they start wPrime. Otherwise your scores are going to be severely skewed. Thought they'd have fixed that by now.

EDIT: I take that back, corwinicre. I've closed wPrime, CPU-Z and everything else. Starting wPrime on my system is not accompanied by a CPU-Z process. /shrugs

 

I was thinking that all default profiles had the maximum set to 100% but now I understand that if the maximum was less than 100% then it would throttle the maximum CPU multiplier back. I'll have to ask the original guy I was helping to have a look at that setting on his computer.

The i5-520UM has a maximum turbo boost multiplier of 12 when both cores are active and that's exactly what you're getting when running wPrime.

Some interesting testing so far. If your CPU is already giving you full turbo boost when fully loaded then there is nothing to gain by using ThrottleStop. If it isn't giving you full turbo boost due to it hitting the turbo TDP/TDC settings then increasing these looks like it is giving a significant boost for some of these CPUs.

When running wPrime with ThrottleStop open, you should be able to gradually increase the TDP/TDC values and watch if the average multiplier reported by ThrottleStop is going up. Once that maxes out then increasing TDP/TDC any further isn't going to get you any more turbo boost.

Thanks for the results so far.

The wPrime fix for me was moving it to either the C: directory or the Program Files directory. I have another C:\Utilities directory and wPrime tries to start in there but craps out with an error.

corwinicre: Your second screen shot at 10W / 30A shows ThrottleStop in Monitoring Mode only so I'm not sure if your CPU was using those values or not.

 

Isn't the power saver max cpu set to 100% by default? I think a lot of us who are attempting to maximize battery life have changed our max cpu under that power plan by setting it to something significantly lower.

 

Here's one with ThrottleStop running with those TDP/TDC values with the CPU at 12x: http://imgur.com/axaUp.png

Do you think raising the TDP/TDC values will keep Turbo Boost on for a longer time when running something intensive for a long time, like gaming?

Edit: Also, where does wPrime store its logs? I didn't see anything where I extracted it.

and btw this should've been said earlier: Thanks for all your work on this, especially considering you don't even own an M11x!

 

OK, finally. i7 640UM @ Stock (133Mhz FSB) - no ThrottleStop
followed by ThrottleStop with settings: TDP @ 20, TDC @ 60.
Thank you so much for your interest and help Unclewebb

EDIT: Higher times with throttlestop off

 

corwinicre: Do you see the FID column in ThrottleStop in your picture. With your TDP/TDC set to default values you are not getting full turbo boost. An odd ball value like 11.48 shows that turbo boost is cycling on and off rapidly so it is not maintaining the full 12.00 turbo multiplier.

You are right on the edge of turbo throttling with this load. A testing program like LinX or Prime95 Small FFTs might put you further over the TDP/TDC limit and the amount of turbo boost you get will drop down further.

When you hit this limit, turbo boost is designed to rapidly cycle on and off hundreds of times a second. The value reported by ThrottleStop is a very accurate calculation of your average multiplier using the Intel recommended method. CPU-Z does not use that method and rounds its data off to the nearest whole number so isn't as accurate at detecting this. You can also try dropping your TDP/TDC values down and you should immediately see less turbo boost when fully loaded.

You're most welcome for this new feature.

TDP/TDC is power consumption. I've seen small increases in power consumption when a CPU heats up so during some gaming, it's possible that this could change slightly and the amount of turbo boost you are getting could start to decrease. All CPUs are different. Yours might be better than average at being able to operate at a low voltage and power consumption level. The actual voltage a CPU gets and its power consumption can vary a little from one CPU to the next even of the exact same model.

ThrottleStop has a log file in the ThrottleStop folder. I'm not sure where or if there is a wPrime log. I've never seen one.

 

Same again, but running on my OC @166Mhz
Huge 10sec+ difference between TS on and off now though, this is good right?

 

Your results look like an 11 or 12 second improvement. Is this a bad thing? It looks like an incredibly good thing to me.

When you overclock your CPU, you are increasing the power consumption of it which means without ThrottleStop, you are hitting the turbo TDP/TDC limit of your CPU sooner. This reduces the amount of turbo boost and the multiplier so your total MHz isn't much different when fully loaded.

Without ThrottleStop, you are being limited by the default TDP/TDC settings. With this new version of ThrottleStop, it's open season and you are getting a lot more turbo boost and performance out of your CPU. I had no idea that there was still so much extra performance hiding in some of these low watt CPUs that has finally been unlocked.

While wPrime is running, have a look at the average multiplier being reported by ThrottleStop. In your two examples, you should be able to easily see a significant increase in the average multiplier.

Edit: Very interesting to see that overclocking your BCLK from 133 to 166 got you no improvement in full load wPrime performance. That's a clear sign that you are hitting the turbo TDP/TDC limit.

 

I wonder how will this impact game performance, if at all...

Anyone try running any tests on CPU-intensive games? I have an R1, or else I'd do it.

 

Is there any value in finding the settings for TDP/TDC where gains flatten? I guess I'm not clear on what the full ramifications are for increasing these settings are. We're essentially telling the system to allow TB to run beyond what's specified for these UM processors. Is there the potential for these settings to be set too high thus creating a over-temp scenario when gaming?

 

On the quad core CPUs like the 920XM and 940XM that can also increase the turbo multipliers; there was a big heat problem when increasing the turbo TDP/TDC limits. Some users were hitting core temperatures of 100C and triggering the Intel thermal throttling mechanism so users had to modify their cooling solutions and/or reduce the TDP/TDC limits.

When you increase these limits with a Dual Core CPU, I don't believe that you will have any temperature problems. You can try setting these values sky high so you are getting as much turbo boost as possible and then go do some gaming and use ThrottleStop to log your CPU core temperatures. If you are having any temperature problems then I guess you will have to pay more attention to the TDP/TDC values. I think you will be able to get away with setting them as high as you like. This new feature also makes SetFSB a lot more useful since you're not going to be as limited by the turbo TDP/TDC wall.

 

quite possibly the highest 3Dmark06 ever achieved on a m11xR1 or R2 thanks to throttle stop thanks again uncleweb im noticing performance increases in SC2 as well in the late game.

 

my own little internal testing with fraps benchmark shows a 41 average FPS over 45seconds during a replay with TS off to 48 average FPS over the same 45 seconds with TS on.

edit: at 1080p with all settings low except textures on ultra.

edit2: LOL this is on SC2

 

No Throttle Stop
9.56 - 1596Mhz Min
9.66 - 1613Mhz Max
with a deviation of about 2Mhz

total time 1024M - 1134.6 seconds
18Minutes
55 - 57C
------------------------------------------------------
Yes Throttle Stop
12.02 - 2007Mhz Min
13.56 - 2264Mhz Max
With a Deviation of about 10Mhz

total time 1024M - 844 Seconds
14Minutes
61 - 67C
------------------------------------------------------
Notes: on my 1st run with the Throttle Stop on (20/60) i got "cheat detected"
FSB - 166.
i7UM - 4gb RAM
------------------------------------------------------

Ok so my conclusion is THIS IS EPIC saved 4 minutes
fantastic results!

on the 32M i save about 10 - 11 secounds.

 

I got a 10 seconds improvement on the 32 test on my M11X R2/8Gb/ with FSB at 154.

 

Greetings. I`m the user that started all this by ing and moaning to a very good friend of mine. Luckily, he knew Kevin

I`d like to thank both of them for their patience and Kevin for his great work.

Here`s some tests I did with v2.87, will post with v2.88 asap.

I`ve included some 3DStudio Max 2011 x64 results also, which, although are not astonishing, might improve even further with the new version.

Here goes the screenshots and logs.

 

Logs now.

Will run the 2.88 version asap.

 

Welcome to NBR Cosmin. I think there are a few happy users here due to you having good connections with my friend burebista.

It's been a long day so I'm heading to bed but I'll check out all your log files tomorrow comparing the two versions to see how things are working out for you.

xSix: When you drop 290 seconds off of your 1024M time, no wonder they think you're cheating!

 

OK guys, I smell love in the air.
Have fun with Kevin and your CPU's.

 

so, using throttlestop and fsb at 154 I ran Battlefield Bad Company 2 on the following settings:

All high, no HBAO, No one left behind level and I got 27/30 fraps, cpu was running at 70 C though


Then I turned off Throttlestop and I got 18-21 max....

 

On some of the other Alienware computers, users were having a problem with the accuracy of fraps after adjusting their BCLK. There are two main timers in a Windows PC. When using SetFSB, these two timers can go out of alignment and the Windows high performance timer won't be able to accurately measure time anymore which can screw up many benchmark programs.

Can someone try running my WinTimerTester program for about 60 seconds.

WinTimerTester.zip

This will compare the two timers and show if they are running in sync or not. If they are not in sync after you overclock the BCLK then you have to be careful with what benchmark programs you are using.

Your results certainly look good niithegoat as long as fraps is accurate.

 

Must be something I`ve been smoking.

A 13 minute gain out of 67 minutes is roughly 20% increase in performance, which correlates with the 10-11 second gain in wprime. So, it did work wonders At the moment I`m running another render test with 2.88, wprime showed an 0.8s improvement, so the render test should be a liiiiitle shorter.

Also I believe I fscked up the 3DMark06, for a 1000 score looks like more the integrated video adapter rather than GT335M. But the CPU score increase from 1350 up to 1678 is more than explicit.

Later edit: been reading the thread and I`ve seen this:

For all the tests Kevin requested I had it set up to maximum performance, manually entering 100% both in Minimum and Maximum Processor States, so what you guessed doesn`t apply

 

Wow nice job guys. I haven't read this whole thread yet but I've always been convinced that there's something wrong with the m11x r2 ... the CPU performance is way too low relative to the size and weight of the notebook and turbo boost is basically non-existent or never works. I'm amazed you're able to address this via software. Will definitely be giving it a try.

 

Ok, so I've run some test on wPrime:

Turbo Boost, No OC: 32M - 36.255 secs
Turbo Boost, No OC, TS 20/60: - 32M - 34.156 secs
Turbo Boost, OC 160mHz: 32M - 38.813 secs
Turbo Boost, OC 160mHz, TS 20/60: 32M - 28.479 secs

That's quite a substantial change.

 

Just want to add, after enabling Throttlestop, I'm experiencing increase in responsiveness of the entire system. Firefox started up in 5 sec before, and starts up in 2 sec now. Similarly for Thunderbird. Who would have thought that just by increasing the TDP and TDC that you can have pretty much like a brand new system. Appreciate all your work, UncleWebb