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

are you still OCed? check the bios

 

thanks guys,
Not OC'd, speedstep enabled. I even went and reset BIOS to the defaults.
Wow i wish i caught an image before rebuilding... i cannot figure this one out!

 

you need to OC to see any noticeable improvements with TS on, were you OCed before you flashed the image?

 

never had the need to OC the FSB with a 40%+ increase in my games benchmark after simply enabling throttlestop and forcing the multiplier.

However, since my rebuild I have tried the OC with throttle stop on, not getting but a 10% increase with the OC, TS ON or OFF so I’m really convinced TS its just not working for me anymore

I was also working on my ASUS G73JH about the time this thread arrived and had recently used TS on that (to stop CPU downlocking on battery). With two system being configured for TS the first time pretty close to each other, I don’t remember exactly what I did to get it working on the m11x, but I pretty much just remember turning it on lol.

Was there anything else I forgot? I doubled TDP/TDC, forced multiplier (10x), set the same on all profiles, and turned it on.
I just tried setting compatibility mode (xp SP3), no go… any suggestions are greatly appreciated. I cant simply reload windows again because my iTunes authorized computers is now maxed and I already “reset all” this year…

 

I noticed little to no improvement without a CPU overclock, and I know others have said the same thing. I wonder if the 40% you thought you saw before was more of a fluke and now you’re operating normal?? Probably not the answer you wanted.

I thought of one thing to double check:
Did you run wprime and compare TS on and off? That would be the definite way to be sure it is doing it’s job. Again with no OC the gains would be minimal, with an OC it can be about 50%.

 

I don't think there's a need to set the multiplier for these UM CPUs, but that's neither here nor there regarding your issue with TS.

Start up TS, make sure it's Turned On and ThrottleStop Enabled shows in the menu. Run a quick 32M wPrime test and tell us what you're seeing your FID values do? Are they the same when running wPrime with TS turned off?

 

At least give us something to go by. A few ThrottleStop screen shots half way through a wPrime run will tell me a lot with and without TS enabled. Include what CPU model you have.

There should be no need for XP compatibility mode. I'm running W7 x64 and ThrottleStop 2.89 is working fine. There should be zero performance difference between it and the 2.87, 2.88 series. 2.85 doesn't have this new feature for the UM processors.

2.89 just makes it easier to set the Force TDP/TDC value. Make sure you don't have a version of wPrime that is starting up CPU-Z in the background and sucking a quarter of your CPU cycles. Check the task manager for that issue. When wPrime is working properly, it is a great bench to measure CPU performance because it is 100% CPU dependent.

Like other users have said, most people are not seeing a 40% performance increase without overclocking. ThrottleStop doesn't really start to pay huge dividends until you crank up the BCLK.

 

Also make sure you're power profile is set to High Performance.

 

So, had my first go at this. Set the FSB to 160 (arbitrarily), booted up, launched TS, then opened WoW which just happens to be what I'm playing mostly right now. And omg. I'm seeing >50% increases in framerates. I was doing 25-30fps outdoors in Northrend (40-50fps indoors), now it's 45-60 outdoors (with the odd spike reaching 80), and indoors is 80-100.

Will try to FRAPS a flightpath to benchmark more scientifically, but this is amazing.

 

Well done, thanks alot.

I was showing off my M11X over the weekend and it was impressive enough without using what I now know about...

This afternoon, after getting home from work, nVidia's driver release created a chain reaction. One that started at [H]ard|OCP for the first news, nVidia's release notes page and then lead me here for conformation that they'd work properly.

Then, all of a sudden I see this other interesting looking thread that had exploded

Running the recent Verde release in conjunction with Throttle-Stop has made this M11X for real.

 

Hi Everyone,

First of all, I'd like to thank those that helped me get wPrime working (MassiveOverkill, djr53 and headphones989). I needed to right-click and Run as Administrator first, that was my issue, then I got hung up again and used the 'hold shift' advice to bypass hardware detection. I feel dumb for the Admin thing, especially having followed this thread from the beginning. +Rep for you all for not making me feel too dumb on that one

My M11xR2 specs are in my Sig; I have the i5.

Here are my wPrime 32M v2.04 results with TS v289 (9 multiplier):

36.5 sec - No OC, No TS
36.2 sec - 166 OC, No TS

28.6 sec - 166 OC, TS (20/60), Force TDP/TDC=16
28.8 sec - 166 OC, TS (20/60), Force TDP/TDC=10
28.8 sec - 166 OC, TS (20/60), Force TDP/TDC= 8


28.9 sec - 166 OC, TS (30/90), Force TDP/TDC=16
28.8 sec - 166 OC, TS (30/90), Force TDP/TDC=10
28.7 sec - 166 OC, TS (30/90), Force TDP/TDC= 8

This is my first OC and I went right to 166MHz for these tests. Yes, I was greedy, but so far so good. Now I would like to do some load tests on the OC to ensure it is stable. What is the simplest program to do this? Thanks.

Also, a big thanks to UncleWebb
Cheers,
Stain

 

Short OC guide:

• Reboot
• Hit F2 when you see the alien head logo
• Once in the BIOS right arrow to the Advanced menu
• Scroll down to CPU Overclock and hit Enter
• Scroll down to 166MHz and hit Enter
• Hit F10 to save settings and reboot
• If you manage to boot to the desktop then you're going to want to stress test your system at the bus frequency you've chosen.
• If you get a BSOD then reboot and try the next lower setting for CPU OC.
• Once you've booted successfully, download and install Prime95.
• Fire up Prime95 and run the In-place large FFTs torture test.
• Go watch a movie and eat dinner as you're going to want to let Prime95 run for at least a few hours.
• If your system doesn't freeze or BSOD then you're probably OK at the setting you've chosen in the BIOS.
• To further test, play a game or two for a few hours, preferably something that will utilize multiple cores. This will get your CPU and GPU going at the same time.
• Rinse and repeat until you find the frequency that's stable for you.

Absolutely agree. Everyone who uses this to play a game that was previously borderline, see drops in the time it takes to encode, etc. should donate whatever they think is fair.

 

Glad you got it working. LinX is going to produce the most stress. You can alternate between LinX and Prime95 if you want. I've had 24 hours Prime95 stable fail within minutes of using LinX however.

 

MO - are you running LinX with the updated libraries from Intel?

 

Do I run Prime95 or LinX with TS running as well? That is the real test I guess.thx

 

Linpack clones put an unrealistic and uneven load on your CPU's.
For Nehalem and latter CPU's prime95 large FFTs pull almost the same heat as Linpack clones but load is constant.
A test (i7 920 @4.2GHz with 1.40V Vcore BIOS/1.35V load cooled by TRUE) to backup my words.

 

I'll check what version it is as soon as I get back on it.

Using LinPack 10.1. LinX version is 0.6.0.1

 

Please do that!
We absolutely need more vids of games as proof.

 

i think video proof isnt much necessary at this point with much words from some of the more reputable posters here. also rather than that wouldnt it be better if you test it on your own games to see the results yourself? this is a boost on the cpu side. so obviously it does not apply in everything. games that are gpu bound should not see much difference. but if you happen to have anything that taxes the cpu a lot you are 100% to see gains between 20-50% depends on case to case

 

thats all you know and I know. enjoy your booze

 

I will also do the same, as soon as i get my new debit card

 

Try 0.6.4. It has some improvements.

 

On the m11x option you added for us i used to be able to set it to 2 but now it wont go under 4. Im not sure why but at the 2 setting it was almost a solid x14 with 4 its 13.97-13.98 solid. This doesnt bother me scince the gains are minimal but i was just wondering if it was somthing i did.

 

There didn't seem to be any significant overall improvements going below 8 so I blocked it at 4 for the minimum in TS 2.89. This setting controls how often the TDP/TDC register needs to be checked and modified. If this happens too frequently then you may see a higher average multiplier but the CPU is wasting so many CPU cycles constantly monitoring this register that you don't gain anything by setting it very low. This setting might also interfere with the cores trying to go and stay in the sleep state which wouldn't be good for battery life. I think 16 is a reasonable compromise for general use and maybe 8 for gaming and perhaps 4 for benching if you see any improvement going this low. Most of the test data posted seems to show somewhere around 7 or 8 is as good as it gets for overall performance.

Thanks everyone for your contributions. $$$$$

I was looking for some data. On games where this tweak shows a huge increase in frame rates, could someone try playing a game with and without ThrottleStop active and post the maximum core temperature in each situation. For a synthetic test you could also do the same with a longer wPrime run like 1024M. An increase in core temperature and a decrease in battery life are the only negatives that I can think of so it would be good if I could post that info in the first post so other users can make up their mind if this tweak is worth it or not.

RealTemp or Core Temp can be used to record the maximum core temperature or any program you like that is accurate. Almost forgot. ThrottleStop shows the maximum temperature too if you hover the mouse over the temperature icon in the system tray if you are using that feature.

RealTemp 3.60
http://www.techpowerup.com/downloads/1872/Real_Temp_3.60.html

Core Temp 0.99.8
http://www.alcpu.com/CoreTemp/

 

With one of the eatlier builds where we were adding the M11x setting in the .ini a value of 6 was locking my i7UM's multiplier at x14 during wP 32M tests. However now I'm having to set it to 4 in order have the multiplier locked. Values above 4 still show some (slight) fluctation.

I think I'll go back to a value of 8 and deal with the slight loss in multiplier.

 

I think for build 2.88.2, I had the code as lean as possible which meant I removed the ability to disable this feature. This is a very powerful feature that is mostly controlled by about 2 lines of C++ code.

The most recent build, TS 2.89, lets you click on the Turn Off button so you can immediately disable this feature if you like. More flexible but probably not quite as efficient. In my opinion, there is no reason to ever disable this feature. You're better off just setting the Force TDP/TDC value to a very high number like 1000 plus if you don't want TDP/TDC constantly monitored and adjusted.

It will take me about 2 seconds of work to compile a more efficient version similar to 2.88.2 which can't be toggled on and off with the Turn On / Turn Off button anymore. I'm more than happy to release a special build for anyone like yourself that has made a contribution to this project. Just send me a PM and I'll send you a link. TS 2.89 SE, aka. Special Edition.

 

Absolutely no need for a special build. To be quite honest I'm not the tweaker I used to be. I generally use moderate settings as opposed to pushing things to their utmost limits. I'm perfectly happy with a x13.98 multiplier. I'm really only ever using TS when playing a CPU bound game that needs it. Otherwise I'm fine with a moderate CPU OC.

 

Too late.

I'm mostly just curious. Check your PM box in about 30 seconds for 2.89 SE.

 

Yea im not too worried about .02. I will do more testing over the week & when i can i will do games as well. Also were was the link to donate? Or is it built in to TS? Thanks alot unclewebb for all your help.
With the results on the R2 with TS & the upcoming fix for PB i finally got the laptop i believed i was getting when i bought it lol *cheers*
Now for some new accessories for this thing anyone see something new?

 

Actually I was mistaken. It was 2.88.3 where I was seeing a rock solid x14 multiplier with M11x=4.

My previously posted 2.88.3 results

I PM'd you a log file dump with 2.89SE.

 

What do you mean more efficient? Just curious. How can a on off button change the efficiency? Dont quite understand

 

Before ThrottleStop updates the TDP/TDC register, it has to do an if statement to determine whether you have ThrottleStop Enabled or in monitoring mode. That was the only difference I could think of so I thought maybe that was causing the multiplier to drop down by 0.02. Obviously it's not important but I was just curious if it was something minor like this causing this.

 

Good work unclewebb. You've finally justified my trade-in of the R1 for an R2 worth the hassle

I tested ThrottleStop on my Core i7 m11x against FFXIV the other night but didn't notice much of an improvement with it enabled. It was around 1 -2 FPS in crowded areas and 4-5 FPS outside. I'm guessing FFXIV is more GPU limited than CPU limited. I would test BFBC2 but that game has Punkbuster issues still so that's out. Single player is not worth testing as it's not that stressful. If someone is still on the 189 drivers please test out multiplayer BFBC2 and report in the results / gains.

I'm using the new 260 modified drivers and I'll test Left 4 Dead 2 and Team Fortress 2 soon as I believe source engine games will show moderate gains. I also can test Shattered Horizon and maybe Street Fighter IV although that game already runs fast. Once I have some results I'll post my hardware configuration settings and report back.

If you feel like being nice you could update your OP with the results that are sent in so folks don't have to dig through 39 pages to look for results.

I noticed with mild GPU and CPU overclocks with Throttlestop enabled that my temperatures went up an additional 10 - 12 degress celcius which is is a little scary. The CPU/GPU is hovering around 85-87 degrees celcius while playing so I might have to get a laptop cooler for this program Perhaps this is what Dell is worried about.

Thanks for your hard work on this, much appreciated.

 

Those temps are well within acceptable limits for both the CPU and GPU. A 1 hour L4D2 session gives me 84C max/ 81C average temps for CPU. I haven't yet checked GPU temps, but they should be fine. Just propping the back end of the notebook up a quarter inch will drop temps 3-5C.

I'd also encourage you to donate a little if you can. The performance improvement that TS brings to the table for the R2 are pretty huge in my book.

 

Ok, I've been doing a little bit of testing with Civ5 (DX9) and COD MW2. I'm going to give Starcraft 2, Farcry 2, Portal and Trine a go as well, eventually... I have to work tomorrow morning though so, let's just start with Civ 5.

*** For all testing, I'm using a registered copy of Fraps alongside CoreTemp 0.99.8 and play for 5 minutes.

Settings:



M11X @ 162MHz, 260.89, No ThrottleStop

Max Temp: 73C
Max FPS: 41
Min FPS: 0
Avg FPS: 28.711
Elapsed Load Time from Main Menu: 1 minute, 42 seconds

M11X @ 162MHz, 260.89, ThrottleStop Enabled ( 30/60, 8)

Max Temp: 79C
Max FPS: 48
Min FPS: 0
Avg FPS: 30.301
Elapsed Load Time from Main Menu: 1 minute, 20 seconds



Ok, so Civ5 doesn't seem to want to work in DX10/11 mode with my 335m and any ideas regarding solving that problem would be appreciated. Barring that, the game looks beautiful on our M11X R2 machines. There really isn't an image that I can provide to give you an example of the increased performance from ThrottleStop here. My game is taking place on a huge map with a full complement of AI opponents and city states. The average frame rates are very similar, but the load times from main menu and between turns (obviously variable) are vastly improved when using the above ThrottleStop settings. This is my favorite PC game right now and this boost has made Civ5 much more playable on the M11X.

Moving on, I fired up COD MW2.

Settings:



M11X @ 162MHz, 260.89, No ThrottleStop

Max Temp: 64C
Max FPS: 68
Min FPS: 41
Avg FPS: 55.830

M11X @ 162MHz, 260.89, ThrottleStop Enabled ( 30/60, 8)

Max Temp: 66C
Max FPS: 94
Min FPS: 43
Avg FPS: 59.890


Big improvements here. For this test, I ran through the training portion at the beginning of the single player campaign as fast as possible... though I'm horrible while using the track-pad. There was a noticeable difference in smoothness with ThrottleStop enabled, though the game was very playable with the above settings utilizing stock performance as well.


Just a little bit of real-life info to go with all the 3DMark06 scores. For reference, I'm hitting just over 8k in 3DMark06 Advanced and posting up wPrime scores just north of 26 seconds for 32M.

Heading to bed, I'll post some more benches later this week.

 

@D3 - Thanks so much for posting this. 4 more fps helps a lot. Some may not agree but if you are setting at 26fps, 4 makes a big difference. Thanks for taking the time to do this. I was starting to get concerned that the m11x community was not willing to do some simple work to show the benefits of performance enhancements like the ones shown by UncleWebb (Throttlestop).

+rep

Thanks, StevenX

 

mmmmmm.......strange i disabled powermizer and enabled throttlestop i still get throttling issues (15~25 FPS) on NFS undercover, strange? The normal gameplay is 30~60 FPS.....

 

DivineAura: How about running a GPU-Z and a ThrottleStop log file while gaming to see where the problem lies.

Sometimes you have to keep things in perspective. Even with ThrottleStop running, an Intel Core i7-640UM is a badly crippled, low power processor and it's not about ultimate performance. Maybe the U should stand for under performing.

With 2 cores active and with ThrottleStop helping out as much as possible, it's going to max out at 1.86 GHz. A very similar Core i7-620M that isn't a low power UM slug is going to max out when both cores are active at 3.06 GHz. The 620M can run over 64% faster than the 640UM when fully loaded.

Someone needs to put the blame on Dell's shoulders and ask them why they don't offer the option of a full speed CPU in this little rig. If this is too powerful for this platform then they could drop all the way down to a Core i3-330M and they would still be able to outperform any member of the UM series. The 640UM model number might give you the allusion that it is a powerful dual core CPU but in reality it is not.

 

Lastnight i was playing games & knowiced when i tabed out to look at temp etc.. that the multi was still going over x14 not like when i run Wprime were its almost solid under load. I saw that there is a spot that shows that core 1 max is x17 still & core2 is x14.
Is there a way to change the setting to set max x14 on core1? This way it never goes to a single core.

 

That's normal behavior you're seeing. When all 4 threads are being /fully/ utilized, such as wP does, then the all 4 threads will have a maximum multiplier set while while TS is enabled. I don't know of any games that are going to max out your CPU.

 

This might be slightly Off topic. and i apologize for it. but there are some things i don't understand about the UM series. I understand that they installed these processors to attempt to maintain a decent amount of battery life and that it is one of the bigger drawbacks to this laptop compared to others. I guess my question is how does it compare to other processors though? I am relatively new to computer hardware, but i have heard ppl say that a pentium 4 2.6ghz is nothing compared to a 1.2 ghz i7. but what about in terms of the UM series? it gets complicated because the um series performs at like with lower power. I am relatively new so now i am wondering if i should have not upgraded to an i7 at all when i purchased this computer in the first place.

ps. loving throttlestop, its definately what this computer absolutely needed. don't worry, i fully plan to donate....just waiting on my next paycheck.

 

I know benchmarks are not perfect at comparing real world performance but it’s interesting that my old I7-720QM scored barely over half what this I7 does in Vantage (using TS of course).

 

When all 4 cores of a Core i7-720QM are active, you are usually over the TDP/TDC limit so there is zero turbo boost. You will be running at the default multiplier which is only 12 for a total of 1.60 GHz. In terms of pure clock speed, that's slower than the 640UM and many 640UM owners are overclocking the BCLK by 20% which also helps. You need to run a few different benchmarks but I'm sure an overclocked 640UM would win a few against the 720QM.

 

Okay quick report.

Specs: R2 i7 w/4GM of RAM, 160GB Intel G2 SSD. 160Mhz FSB, Default GPU clocks. Geforce 260.89 drivers. Using FRAPS to measure.

Shattered Horizon 12-14 players

Before:

14-21 FPS.

After:

17-24 FPS.

Felt smoother with it enabled. When the frames are this low to begin with it really helps.

Team Fortress 2 24 player server:

Before:

35-70 FPS

After

45-90 FPS.

Big boost in this source game as expected. The game runs fine already without tstop but in large open areas with lots of players fighting you can really tell there's a boost in FPS.

Please note I'm just looking at the FRAPS fps while playing so my numbers could be off a little. I'll try to get some more games done if I get any more time.

 

I must be doing something wrong here...

I ran wPrime v2.04 multiple times and my fastest times have all been at stock (33-34 seconds, with no overclock and no ThrottleStop, and with Turbo Boost enabled)

I tried multiple different OC settings (150, 158, 160, 164), and on each I tried with and without ThrottleStop and all hovered around 35-38 seconds)

I'm pretty sure I have the right settings in ThrottleStop.
Clock Modulation on - 100%
Chipset Clock Mod on - 100%
Set Multiplier on - 10
Force TDP/TDC - 8

The CPU stays at max and C0% stays at 99.8 on all while running the test, so I'm pretty sure the settings are right. Anyone know why my times are getting slower?

 

You missed the most important setting. You need to right click on the TS window and Choose Turbo Power Limits. Set each of those to something high like 100/100. What this is doing is raising the limits for TB and netting us some significant gains in performance when our CPU is fully loaded or close to it.

Also you need to uncheck Clock Modulation and Chipset Clock Mod as neither of those settings have anything to do with our UM processors. I haven't seen that the Set Multiplier setting does anything for us either, so uncheck that too.

 

Aerotype: I agree with slickie88. When you start increasing the TDP/TDC, I think your wPrime times should start to fall and be similar to what other users have posted.

Depending on how you have Windows set up, you might be OK with Set Multiplier unchecked. I usually just max this setting and check it to be on the safe side. You can do some testing to see if it is needed or not.

 

FYI I've set Real Temp to run in Windows XP compatibility mode and so far RT adn TS are playing nice together. I'll update if I see any future dll errors.

 

Haha mr webb posted 4 times,
I will check my FPS in Cod4 later today,
This isn't a high CPU game but i will check it out

 

Uncleweb - you mean to tell me a the I7-620um with 4 threads is going to beat the I7-720qm with 8? I think not. If you disable 2 cores on the I7-720qm to make them equal the 720 "should" clock up to 2.4ghz on 2 cores and 4 threads should it not? I think it's probably pretty close but also the 720 will more a slight cache advantage as well, 3MB more per clock cycle. It wont equal much in real performance but it a slight edge. I will be happy to run a comparitive bench against the 620um when I get my 720 installed either tonight or this weekend. I am curious about it myself. Thanks, StevenX