AW M15x Throttling Issue Investigation - Stock clocks and overclocked.

I tried fallout3, indeed it helps, but it doesn't 2k10 and gta4...

 

Are there any i7 performance laptops which actually are reliable in the world of laptops which don't have problems? I mean, we are paying a lot, it shouldn't have problems when we just opened up the laptop. Any recommendations? (which is actually available in the UK)

Nearly every single laptop seems to have issues nowadays..

 

Clevo's. 10char

 

In my current gaming/applications I have had no problems with my M15x

 

same here no issues

I do acknowledge its there but i could care less until it happens to me ...... if i worried about every thing that was wrong with every computer i had, well i would of smashed them over my knee a long time ago

 

Something is definitely wrong then.

Even with the power sucking 920XM in my system and the more pronounced reported dips via GPU-Z, my fps in dalaran are 48-60fps at 1600x900 with everything to max, 1x AA/AS and shadows to min.

 

LOL, you are enjoying this WAY too much methinks.

 

I just keep reminding my self who can overclock

( he has the non cut out bios but still)

 

I have a copy of both the overclockable and non-overclockable BIOS, but I flashed the non-overclockable BIOS onto my notebook because I don't really need the CPU horsepower. The same thing goes with my GPU; I know that it can surpass 680/1700/1000, especially after I mod the heatsink, but I will lock it at those clocks because I don't need that much power.

 

i understand what you mean its why i dont have or want a quad core

waste not, want not

 

The i7 is quite something though. I am able to play my dual-threaded PS2 games much smoother at 3.06GHz compared to my previous QX9300 at 2.93GHz.

 

you have the overclockable BIOS for I7? I have a 920xm and I really want to overclock it! Would you share your overclockable bios with us please! Thank you so much!!!

 

He owns a Clevo, not an M15x.

 

I have finished some testing on my M15x using Computer Chess as a "Real World" application to stress the CPU and check for Throttling. To do so I have ran an Infinite Analysis of a Chess position, confirming that CPU utilization is at 100% on all Cores and sustained over a 30 Minute Period. So far I have not seen no GPU throttling. This may be due to the fact that minimal demands are placed on the GPU. I have tested with the Deep Fritz 11 GUI 3D Chess Table running but no evidence of throttling noted. Details as follows;

System Spec;
Alienware M15x
BIOS Phoenix A01
1600 x 900 Display
Intel i7-720QM CPU
4GB DDR3 1066
1GB GDDR3 260M
500GB 7200 RPM SATA HDD
DVD±RW drive

Laptop is Stock with the Exception of Mcafee completely removed and running Kaspersky 2010 AntiVirus, Windows Defender & Firewall and any applicable Windows updates.

Software;
Windows 7 Home Premium 64 Bit fully updated
Chess Assistant 10 with Rybka 3 Multiprocessor version 64 Bit, 256MB Hash.
Deep Frtiz 11
Aquarium 2010
NVidia Driver v186.47 (Stock)

Tools;
GPU-Z 0.3.8 - GPU Clocks stay at 550/950/1350 on my GTX260M (Driver 186.47)
Real Temp 3.40 - CPU Pegged at 100% 1728.98 MHz, 133.00 x 13.0, Temps Peaking at 74 Degree's Celsius
Windows Task Manager - Confirms 8 Cores (4 Physical Cores w/HT) all at 100%
Everest v5.30.1900 - Used as a Check against results from GPU-Z, RealTemp, everything confirmed.

Using 2.36GB RAM

Going to try some different Tests however so far so good, also as a side note computer chess, particularly analysis, easily pushes all cores to 100% (Assuming a Multi-Core version of the Chess Engine)

 

hey guy, i found out the method to solve the problem, go get a 220W or a 250W AC adaptor will solve the problem

 

Apparently you had not read the previous thread - Joker already tried that and it yielded no difference. Though, he was trying the extreme scenario. The persons posting the other throttling with <100% CPU utilization may still benefit from it.

 

This might help all that are using WoW to test. Don't know if u guys know this but wow by default is set to use only two cores.
To utilize more cores there is a client side setting.

If you are using an multiple cores you can set the ProcessAffinity in the /config. file.
SET processAffinityMask "XX"

Example if you have a i7 and you want it to use all cores logical and physical.
Meaning all 8 threads. You put the following:

SET processAffinityMask "255"

Another setting often used by players for i7 procs is:

SET processAffinityMask "85"
I believe in this case 85 will just use the physical cores.
I will research and get solid answers.

I hope this helps someone.

 

Does any of the examples of throttling been using a 720qm? Perhpas it happens only on the higher end ones like 820qm and 920xm ... Just and idea . as I can assume those procs use more power. Also how does ram affect power? Would a single 4gb chip take more power then a 2gb chip?

Just food for thought.

Regards

 

I have a 720 and was able to duplicate jokers results. they are very extreme tests i havent had it happen real world at all yet but im sure it happens. Now if theyll fix the headphone output to that would be great. AND stop using seagate harddrives

 

As we talked about over PM and in this thread, the GPU has to be utilized to a certain extent to see throttling occur. The throttling problem with the M15x seems to be related to a power threshold. Once it's crossed, the GPU is scaled back.

If anyone wants to see it work in real time, download ThrotleStop and while Furmark + Prime95 are running, keep lowering your CPU multiplier until you see the GPU no longer throttles. That is the limit of your system. In the real world, this is happening in games where there is a lot of activity which utilizes at least 1 core to 100% and pushes the GPU as well.

Chess likely does not push the GPU at all so pushing the CPU alone won't induce any throttling. I'd suggest that if people are looking for real world results, make sure the GPU is sufficiently pushed while doing so because that is the basis of this throttling discussion.

Thanks for your input snarl.

 

The 240W power brick does not help at all even under regular clocks. This issue is definitely a bios one or a limitation of the system itself placed by Dell. I'm willing to bet it's the former and not the latter.

 

I found that utilizing the Blueray drive resulted in my system throttling with less load on the CPU/GPU. The Blueray drive appears to consume about half as much power as the CPU which may explain why.

 

If the mobility 5800 series does end up consuming less power than the GTX 260M and Dell combines it with a dual core arrandale chip, the throttling probably won't happen at stock clocks anymore. I believe Dell is capable of correcting this flaw and they should issue a bios fix. They did it for their latitude series after someone had to go through the effort of putting out a 56 page pdf file detailing the issue. Trouble is, if they do issue a corrected bios, it could very well be that the 150W PSU they provide cannot hold that load and they would have to issue a bigger PSU for existing customers like they may have to for the SXPS 16.

 

Interesting stuff. I wonder how much Dell engineers are getting paid to "research" this stuff? Hopefully you guys have done enough to find a smoking gun, for lack of a better term.

 

It might not even be fixable with a bios update, which would be the worst case scenario for us M15X owners (since it would be unlikely to get fixed)

 

I am split between hoping for a ray of sunshine in January with introduction of Arrandale - or just purchasing an Studio XPS desktop

 

has anyone tried setting maximum processor state to 95-98% in windows power setting and then running prime95+furmark?

 

Good idea.

I tested with my 920 and it starts throttling when I set min + max of processor state for anything above 85%. At 85% it runs without throttling

Picture

 

Hmm, what about undervolting(as described here- http://forum.notebookreview.com/showthread.php?t=235824&highlight=undervolting), can u do that on an i7 and has anyone tried it?
it ought to solve the problem as this seems to be some sort of power problem, and even if it doesnt solve it entirely, it'll let u set the max processor state higher, nearer to 95-100

 

I can understand the confusion. There are two behaviors here, both related to downclocking the GPU, and they are being confused.

Here is the part of the document engineering pointed me to, relevant parts in bold:

When the notebook is plugged into a wall outlet running on AC power, users want full performance because battery life is not a concern. PowerMizer 8.0 recognizes when a machine is running on AC power and allows the GPU to run the clocks up to the maximum if an application requires it. PowerMizer continues to monitor GPU use and still conserves power when the GPU is not being taxed, resulting in lower electricity costs and sparing the environment.

I was told that powermizer will dynamically adjust clock settings during a GPU load, and the result would appear to be fluctuations in your graphs. Exactly how power management works on the M15x isn't really a matter I can disclose.

This is completely different than the cyclical fluctuations in Joker's graphs, which are still being investigated.

I fail to understand why Joker is calling for an internet campaign to say we aren't investigating the issue when nothing could be further from the truth. We are looking into the behavior he pointed out, but I want to make sure people don't get the false impression that fluctuations in the GPU clock frequency are abnormal.

 

I had to go to 83% to get it stable on my 820. Throttlestop was stable at 10x multiplier. Out of curiosity I ran a couple 3D Mark Vantage tests for comparison (Physx off).

Full speed:
CPU = 11,556

83% CPU Limit:
CPU = 7,598

CPU 100% and Throttlestop multiplier 10x:
CPU = 7,646

So at stable rates the CPU performance is reduced by about 34%.

 

Any chance that mrZombieBunny redo his tests with Powermizer disabled just to rule it out as a cause? Or anyone with these problems at that.

 

I was wondering what sites have paid attention to this? Any articles yet? If there is anything please post it.

ThX

 

Have you addressed ANY of what has been posted here? All we're hearing so far is:

1. Hey guys we know it's there but don't count on a bios fix. I'll quote you:

Telling us you're continuing to investigate it but not to count on a fix seems like a cop out. I' call it like it is and others here are reacting the same way I am so it's hardly a "why is Joker out to get poor Dell" situation. Your engineers aren't morons, its pretty obvious they designed the system to throttle at a certain power threshold. If that is not the case, present us with a plausible reason why this throttling occurs because all the tests that have been run so far point to a power issue (via bios by design or a system design flaw).

2. Hey guys this never happens in the real world..if it does PROVE IT. So users proved it and what do we get?:

Ok as far fetched as this is, I'll bite. Let's have Wattos, Arkhias, Electrosoft and MrZombieBunny re-test their findings except this time with powermizer disabled (which can be done via the control panel by the way). If they get the same results, you have to agree not to feed us this powermizer line anymore. If it fixes their problems, then we can chalk it up to powermizer for those cases and continue to focus on the throttling that occurs when the CPU/GPU are under load. Deal?

P.S. I haven't called for any internet campaign against Dell. What I have asked is for other tech sites to examine the issue (which Dell openly acknowledges) that's been presented and to come up with their own findings.

 

How to turn off the nvidia PowerMizer? I can not find it in nvidia control panel

 

yeah seconded - i can't work out how to turn it off. the reg hack usually described doesn't work for the newer drivers.
mine is already set to 'prefer max performance' in control panel

 

***DISCLAIMER: I TAKE NO RESPONSIBILITY SHOULD ANY DAMAGE OCCUR TO YOUR SYSTEM. TRY THIS AT YOUR OWN RISK!***

IMPORTANT UPDATE:


I think I've stumbled upon a temporary fix that will enable everyone with an M15x to run the system at max clocks under max CPU/GPU load. I should have tested this earlier but I ruled it out because I ran the same test with the 150W adapter and not the M17x adapter I also have.

The fix has actually been documented on the first page and the following has to be done:

1. Configure RivaTuner (use the latest version from Guru3d.com) exactly how I outlined in the first post.
2. Use your existing 150W M15x adapter.

Now run all the stress tests you want, the system will not throttle.

I will be editing this post with a video I filmed using my iPhone. If this proves to be a stable fix, then I think this will strongly show that this is a bios limit and not a system design issue.

***DISCLAIMER: I TAKE NO RESPONSIBILITY SHOULD ANY DAMAGE OCCUR TO YOUR SYSTEM. TRY THIS AT YOUR OWN RISK!***

Video:

<param name="movie" value="http://www.youtube.com/v/YcTDGLT2vhc&hl=en&rel=0&ap=%2526fmt%3D18″"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/YcTDGLT2vhc&hl=en&rel=0&ap=%2526fmt%3D18″" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width='425' height="344"></embed></object>

 

Just Curious but according to dell's website the M17x powersupply is 210w not 240w was looking to order one to have a spare just wondering were you got one.

 

It's a 210w/240w switchable supply. Under the bios it reads as 240w so that's why I refer to it as 240.

 

cool just double checking before i order one.

Just ordered one for next business day shipping hopefully it will get here quick. See if this is a working fix if so dell owes us all new power supplies! and a Bios fix

 

Hey John should hand out free supplies to all of us. What do you say John?

 

BTW guys, try this on your 150W adapters and see if it works. On mine it locked up after 15 seconds so I didn't try again until today with the M17x adapter.

Oh and once again: ***DISCLAIMER: I TAKE NO RESPONSIBILITY SHOULD ANY DAMAGE OCCUR TO YOUR SYSTEM. TRY THIS AT YOUR OWN RISK!***

 

Made changes with rivatuner and running furmark and prime no throttle at all been running solid for 20 minutes and counting on my 150watt

 

Awesome! I'm guessing there's a variation between each adapter. My 150W couldn't handle it and the system froze. It seems yours is doing fine so far. We need someone with a kill-a-watt to test how many watts are being pulled during this test. If nobody steps up, I'll drive to ace hardware tomorrow and see if i can procure one.

Edit: I noticed your furmark was being run with different settings. Make sure you set it how I described so there's no variation between results:

Resolution: Not really important but lets set 1600x900 as a baseline.
MSAA: 16x
Checkmark the following: Stability test, Xtreme Burn, Post Processing

Try the above settings and re-run your test with RivaTuner.

 

have one of those on order too , im still getting the M17's PSU to be safe. Now to fix my Headphone issue and get dell to credit me some money for needing a bigger powersupply , having to replace my crappy seagate harddrive and having to buy another sound card. My old gateway FX series gamer was easy to deal with than this..... and i sold that when i got this!

 

Well at least we have a fix now. I'm going to be creating profiles with RivaTuner, one for gaming and the other for regular use. It sucks that we have to hack our way to a stable GPU clock but at least that option is now there.

 

running now at those settings

 

Here we go again works fine tell you what though my powersupply is mighty warm thank got its cold here in the chicago area my room is 67 degrees with the heat on and its 25 outside.

 

Thanks, +rep for confirming. You might want to resize the picture though.

 

Joker - I apologize if I took your callout as an internet campaign against Dell if that's not the way you meant it. I know tone is hard to convey in text.

What you consider a "cop out" is me not making any promises when testing hasn't been completed. I'm not going to unveil an outcome when I haven't received all the facts, and when those who designed this system are still investigating.

Even if you refuse to believe it, the fluctuations you see in those game screenshots are certainly the result of power management, and this is according to those who designed this system. I'm not trying to sound defensive, but if you really think that's me "feeding you a line", I'm really not sure what more I can say to you about this, as you've obviously already made up your mind.

All that aside, I must say this:

I don't recommend using a power supply not intended for this system.

I'm glad you gave a disclaimer when offering your suggestion. I just felt it necessary to reiterate how risky it is.