AW M15x Throttling Issue Investigation - Stock clocks and overclocked.

oh yes... they did just do that

ohwell... I'm sure that'll change...

 

Absolutely wrong! The GPU doesn't need to be at 100% to throttle when the CPU hits 100% load. In fact, the GPU can be sitting at 50% load and it will STILL throttle because the active state of the CPU hits 100%. I've documented this fact countless times. If the Dell engineers can't understand this simple concept, then Dell should fire them and look for some competent engineers.

NO other gaming system that is in competition with Alienware has this problem. Bottomline, AW should NOT have this issue with a premium notebook.This will undeniably make Clevo happy because I'm sure more and more people will opt for their notebooks over a throttle plagued AW with engineers that refuse to acknowledge the problem.

I beg to differ, the Clevo systems offer superior performance because they do not throttle. They do no not have to worry about their systems buckling under pressure. But to be honest, I expected this type of excuse from Dell and I will pass this on to Jared and company at Anandtech and other sites so the issue can be examined in more detail.

 

If I wanted a pretty laptop, I'd have purchased a macbook. This excuse reminds me of guys that put big mufflers and ugly neon lights on their puny civic four cylinder engines and think they have a performance car. Pretty sad that the AW mentality has come down to this.

 

Or a Clevo.

 

Well to fair, the M15x's hotness is one of its main attractions

 

I didn't come here to participate in an argument. I was trying to be jovial.

But anyway, I haven't seen any evidence that these tests reflect real-world scenarios, nor have I seen evidence that other manufacturers with the same chipsets perform any differently under the same circumstances. If you guys have anything or if I missed it, please let me know. I'd be happy to take it back to our engineers. I'm here to do what I can to help.

EDIT: As to the mentality comment you edited in, I'm very sorry if I offended you, as that wasn't my intent. I'm only here to help.

 

Please read the first post in the thread, Sager 8690 owners have run the same outlined tests with no throttling. I will ask one of the several Clevo owners I know to take screenshots so that I can add it to the first post.

I appreciate you trying to be jovial, but for consumers that spend $2000+ on a supposed performance notebook, it's difficult to take a serious issue and laugh about it. Other users have posted real world scenarios but you insisted they contact tech support.

The fact that the GPU throttles when the CPU active state hits 100% means this is not a gaming system designed to last more than a year. What is Dell's excuse going to be when more quad core optimized games hit the market and push the CPU to peak loads more often? "Sorry guys, but the M15x should not be pushed like that, did you think you were buying a gaming system?" See why we aren't in a jovial mood?

Again, I appreciate the fact that you took the time to contact Dell engineering about this and keep us informed. My disappointment lies with the response they gave.

 

Hey all, just popping over from the SXPS16 thread.

I find it promising that Dell is at least being communicative with the owners here, even if it is through middlemen.

However, given the level of technical discussion which has been reached (at least over in our thread), with overwriting the values of CPU registers in realtime to avoid BIOS-imposed limits.

By using a higher wattage adapter plus the program (called ThrottleStop, which was custom made for us by the guy who makes RealTemp) we've seen a complete elimination of throttling in ALL situations, plus a 3Dmark increase of 4000 points. On the 90W, we were getting around 6000, and on the 130 + ThrottleStop we're getting 10000.

I find it ludicrous that Dell engineers (or should I say "engineers"?) have made the determination that it does not happen in real-world scenarios. People would have never NOTICED it if it were not for it happening in the real-world. People were experiencing stutter and low framerates in games such as Crysis, which caused them to investigate further. Then they discovered that the laptop got a higher 3Dmark score when UNPLUGGED than PLUGGED IN.

This is something that has been empirically tested in a variety of situations, and proved consistently in all of them. Dell has "system captured" two 1645s from owners in the other thread, and we still have heard nothing as to what they've found. I'm unsure why they need a specific laptop as they can just pick one up off the assembly line and get the same results, but that's the least of my issues with this whole thing.

We've IDENTIFIED the problem. It's Clock Modulation bringing the laptop to its knees, due to a throttling policy put in place so that the laptop never draws more than 90W. Some configurations of the 1645 draw 140W at full load, unthrottled. Its not as if it were throttling because the laptop is too hot, either. The laptop remains extremely cool when throttled, and only gets hot when actually pushed to its limits.

We've been promised a BIOS update today by various techs that is touted to solve our problems. LionellAtDell has also said a Direct2Dell post regarding our issue will go up today -- or possibly next Tuesday, we're not sure. Either way, many of the people in the other thread, myself included, are extremely unhappy with the level of contact and service Dell has provided, and are strongly contemplating legal action if this issue is not remedied in a timely manner, or if good-faith efforts are not continually shown to take us and the problem we are experiencing seriously.

 

QFT!! alienware looks like a dark road, that at first i wanted to go down in, now i want to avoid at all costs....WHY ALIENWARE??!? u had a near perfect laptop, now its ruined.

and....WTH...10K w/ a studio XPS 1645? screenz plz!

 

Sorry, wasn't me that did the 3dmark testing, but I can quote you the exact numbers that he posted:

Compared to

Big difference, no?

 

i was able to duplicate 5150Joker's results but you really have to pound this unit to get that to happen i havent had a single game or even when running folding on cpu and gpu at the same time it didnt happen. but none the less it is disturbing to see from a laptop of this caliber. Luck for me that i havent noticed any slow down in any games yet! still would like to see this fixed. There is no reason at all to throttle this unit when folding on cpu and gpu the gpu never went over 50c and the cpu never over 54c and those were 100%loads.

 

o that is CPU score from vantage, thought u were talking about the overall socre in 06, cuz im pretty sure u cant raise 4k points just by a CPU, its mostly GPU dependant and the 4650 struggles there.

 

The 1645 has the 4670, btw.

 

actually what is the problem? when i was trying to update my bios, this problem occur!

 

problem occur when i did the test

 

at that prompt type DFLASH EC_W670.096 to finish the process.

 

Thanks for dropping by. I just want to assure you that a fix for the 1645 is still coming. I also wanted to assure you that I am a Dell employee, not an agency or middleman. I'm also the Dell employee authorized to speak here regarding this issue.

I do want to make this very clear to everyone: the issues that XPS 1645 owners are experiencing are in no way related to the behavior you're seeing with the M15x. I can not stress this enough.

How are they different? Well, I'm glad you asked. They have different causes. They also occur in different scenarios, the 1645 issue occuring at normal CPU loads when the system is plugged in, the M15x throttling occurring only when all 4 cores experience a sustained 100% utilization, not a real world usage scenario.

As many have noted on this thread, this behavior is not seen when playing any games... you have to put the system into a situation where performance would be degraded anyway, and a quad core i7 takes a LOT of punishment before sustaining 100% utilization on all 4 cores.

 

@ putochip0123: No Shirt, No Shoes, No Service

 

First off thanks for the input, I have to admit I have been pleased with the gaming performance on my M15x so far i.e. LOTRO, L4D2, CODMW2 etc.. (Loading up Dragon Age next) I am also just finishing off loading up all of my Chess Engines/Tablebases etc.. and will try that to see what happens with computer chess. As an FYI Computer Chess typically pushes the cores to 100% i.e. Deep Rybka 3, Deep Fritz 11, HIARCs 12.1 etc.. particularly during analysis (I use Chessbase 10, Deep Fritz 11 and Aquarium/Rybka) this is one of the reasons I went with the M15x and i7 720QM (My First Alienware).

 

"As many have noted on this thread, this behavior is not seen when playing any games... you have to put the system into a situation where performance would be degraded anyway, and a quad core i7 takes a LOT of punishment before sustaining 100% utilization on all 4 cores."

As others have pointed out, it can be seen when multi-instancing MMOs. While I haven't done this recently, it isn't outside the realm of reasonable expectations from a laptop of this caliber. I also appeared to have throttling when playing City of Villains while listening to a CD when vsync was off. I did not pay attention to the CPU utilization, but I do know the GPU would have been near 100% and the power draw was bouncing around 100-150W from the wall.

Thank you for keeping us in the loop!

 

@ snarl -

That's an interesting scenario. I've used Deep Fritz too, and yes, it does push cores a bit. I don't think I've ever seen sustained 100% utilization across all 4 cores with it though... pegging 100% and dropping is what I've seen. Sustained 100% utilization isn't even really seen with malware...

 

No problem!

I have also multi-instanced MMOs on this box... notably EVE Online... had 3 windowed instances running and was getting 45+ FPS on each. I didn't have any problems.

EDIT: Yeah, if you're seeing performance issues under the CoH scenario you describe, I would strongly suspect something else. This box should scream under those circumstances.

 

where can i go to download the furmark is there a link anyone has so i know im downloading the right program

Thank you

 

John, thanks, I think that is the most info we have extracted from anybody at Dell concerning the SXPS 16 situation. Any idea when a fix is forthcoming?

 

JohnBatDell
thanks for the posts and the clarification that (a) Dell is working on a solution (b) someone, you, who is part of that process is taking time and effort to correspond with users, and (c) that the issues with the M15 and the xps are different, how and why.
Now, can we have that fix today and massive free upgrades for everyone who has posted on this thread (well at least those that have done all the diagnostic stuff anyways)?

 

Joker added links in the first post.

 

I like your style. Blunt. Honest.

But no... a fix for the XPS behavior today and massive free upgrades aren't in the cards.

 

Wait, so this CAN'T be fixed with a BIOS update? Is it at least stable?

I *DO* run my CPU and GPU at 100% on my current notebook all the time, through Folding. Heck, Blu Ray hits my dual core at maybe 67% (one core fully used). Some current games, and probably most future games (and other intensive apps) will probably hit it that hard too :-/

 

You are correct that analysis will not constantly peg 100% usage in all cases however in many it does. For example it's not uncommon running IDea Analysis in Aquarium with Rybka 3 or in the Chessbase/Deep Fritz 11 GUI with Infinite Analysis on a specific move to hold 100%. A general Analysis will peg to 100% then tend to drop off as the evaluation moves to the next move, ply/depth will effect this. Either way I'll run some tests over the weekend to see if I can get a better handle on this as I have yet to truly monitor CPU usage against CPU Multipliers/Clock when analyzing a game so I'm basically just making noise . All my game tests so far i.e. LOTRO etc.. are working great and I've been very happy with performance so far. Based on this you appear to be correct that real world gaming as it currently stands will not trigger this throttle issue (Computer Chess to be determined).

Confirmed 12-17-2009 - Infinite Analysis of a Move will hold & Peg CPU to 100% using Chessbase 10 and the Deep Fritz 11 GUI and the Deep Fritz 11 and Rybka 3 Engines. As a point of reference I will use the Deep Fritz 11 GUI,Deep Fritz 11 Engine with a 3D Table to test my M15x this weekend.

 

It is likely that it is fixable through a BIOS update. As John however said, Dell engineers have considered it too risky to fix as they apparently dont know what will happen if they fix it. On the other hand, there is still a chance this is hardware related, and the fix would be too costly for dell as that would mean that every M15x owner is eligible for this hardware fix

But there is yet no "real" case of throttling while using your computer. I ran today 4 instances of UT3 and there was no throttling

 

Thanks for the info-but from what I've read in this thread, there already are real world cases, which is how this was discovered. Crysis apparently does it, presumably Folding @ Home would (if run the way I would run it at least), and more and more games will start triggering it if they don't now. (Not to mention anything else intensive, like video encoding, etc.)

 

Wait, so Dell is NOT working on a fix? Everyday 100% utilization of the Core i7 is right around the corner once more multi-threaded games come out.

 

Yeah, that's what Wattos said. As of right now Dell's not going to fix it
Besides some of the games listed here that can trigger it, I bet it would trigger if Folding @ Home is being used on it properly (1-2SMP clients + GPU client).

And the other system I'm most interested in (Asus' G51j) isn't stable...so...yay.

Since the GPU is only guaranteed to work at half speed, maybe Dell should start charging half price on the M15x?

 

I know this is off topic but still a concern for me.. Sup with the touch pad on this thing? Is it me or is there major probs.. The movment isnt smooth at all... the only way I can compare it is ... its running at a low frame rate only when the TP is used. If I use a mouse its normal and smooth as glass. I have old crappy gateways where my mouse movements are smooth. This is driving me nuts to be honest. My only solution so far is to remove the drivers for it and that only makes it better not completely solves it.

This touchpad I have had replaced and still the same.. plus if I install the normal synaptics drivers it gets worse. So I have to use Windows drivers to lessen the problem.

Seems this should be a focused issue as well but certainly the hardware throttling is a giant concern.

I have compared this to 6 other computers and for sure this isnt a quality to be expected in a premium laptop.

Part of me thinks its the surface of the touchpad.. part of me thinks its the hardware/drivers either way .. its horrible .. someone please help .lol

I wish there was another touchpad option or something..

I really have strongly considered what I could do to fix this or use another laptops touchpad but obviously that would be a pain and I feel I shouldnt have to do that... If anyone has this same problem please let me know.

Thanks

 

Not true. This behavior occurs as soon as the CPU cores hit a 100% peak, it does not have to be sustained. The fact that Dell is aware that the system cannot sustain peak CPU performance should raise red flags for anyone considering purchasing the system. Dell is working under the assumption that nobody will run multiple instances of a game (e.g. MMORPG games) or that a title or application available today will not push the CPU to 100% peak.

I do not have an extensive enough library of games to test for the issue and Dell knows this. I am encouraging all M15X owners that read this to conduct your own extensive tests. Keep GPU-z open with the logging function turned on, use RealTemp to monitor your CPU's active state with logging enabled. Play your favorite games and/or applications and if you note throttling, please check your logs and report back here. If you have a confirmed case of throttling due to peak CPU loads, please PM me.

Since Dell has taken a stance of not wanting to correct the issue, its up to users at this point to actively begin documenting cases. Once we have sufficient cases that have been tested, we can present our findings to bigger websites such as Anandtech, HardOCP and others.

Thanks everyone.

Edit: Keep in mind, Dell is fully aware of this problem and I'm 100% sure the engineers know how to correct it. If they do not, they should hire unclewebb (creator of realtemp). Bottom line to those that do not fully understand the implication of this problem: If more quad core optimized games are released in the near future that push the i7 towards 100% usage AT ANY TIME (not just sustained), your GPU will throttle to ~50% performance. Do you want a future proof gaming system that is advertised as such or do you want to accept Dell's lame excuse and throw your money away? The synthetic tests prove there is a problem and Dell acknowledges it.

 

@wolfpup - Our engineers have found that folding@home runs the cores at 90-95% and doesn't trigger this behavior.

@snarl - Please let us know what you find. If I'm wrong about those programs, I need to get that information.

 

5150Joker, I think you said you'd already contacted Anandtech, right? I'd hope they'd run this as an article right away :-/

That Core i7/USB issue should be a huge issue on tech sites too. I'm shocked I hadn't heard of it until today.

 

What configuration though? Normally with a Core i7/GPU setupt, you'd run 2 SMP clients and a GPU client. I'd be shocked if one SMP client doesn't trigger it, but two it seems impossible it wouldn't get triggered.

Honestly I REALLY would like to just order this (today even, it's my birthday! ), but between this and the apparently universal Core i7/USB flaw...

 

What's the USB issue? I was thinking of buying the i7 M15x today. Do you think I should get something else? Is it safe to buy this product?

 

I hope someone there at Dell addresses this throttling issue as well as the touchpad... This is crazyness.

 

I have not contacted Anandtech or any other major website yet. I'm still counting on Dell to resolve this issue instead of dancing around it and making up excuses.

Furthermore, I need users like you and others to begin testing and logging your experiences - which games you do not see throttling and which you do. If you do note throttling, check your logs and confirm it (using the programs I mentioned in the previous post [RealTemp and GPU-Z]). Post back in this thread and PM me with throttling results + logs so that they can be documented in the first post.

I hope nobody buys into the excuse that a system upgrade should be warranted if the CPU peaks at 100%, that is entirely laughable.

 

I will be willing to run tests

 

As I have said before, our tests have shown that when all 4 cores reach 100% utilization, you will see a throttling of the GPU... I have confirmed this with our engineering department, and rest assured the tools they use to test this are more advanced than those being used in this forum.

Logically, for this to be a performance problem, that state must be sustained. I have submitted that in this scenario, were there no GPU throttling, the system would be overtaxed anyway, and you would see performance hits, as well as a strained heat solution. Full sustained 100% CPU utilization isn't a desirable state for any computer. I'd further submit if you're pushing a computer that hard, you're using the wrong computer for the job you're trying to accomplish.

I have also submitted that this machine eats computer games for breakfast, and that full 4 core utilization isn't a real world scenario, nor is it a reasonable expectation of any device, unless perhaps the device is being cooled by liquid nitrogen.

I've run multiple instances of an MMO on this machine with no problems whatsoever. This machine will perform. I think that's the most important thing to consider when you're looking at purchasing any machine.

EDIT: Do bear in mind that I'm willing to examine any real world scenario where this behavior is triggered. That's really the bottom line... if it doesn't happen in reality, and won't happen in a reasonable situation, then is it really something you're going to see beyond artificial benchmarks?

 

Much faster to be exact. The M15x and other GTX 260M based systems are pretty far off to be calling the greatest of anything... or even "All Powerful" anymore.

http://anandtech.com/mobile/showdoc.aspx?i=3689&p=9

 

I just found out about it today-apparently it's common for all Core i7 models to have issues with some USB devices, like the iPod/phone/iTunes. It's not specific to the M15x, but rather a bug in either the Intel PM55 chipset, or Intel's drivers for it. (Which is weird and unfortunate, given Intel's chipsets are normally super compatible since half the technology on them was invented by them to begin with).

Personally between those two issues I'm not very comfortable buying one...which stinks, because it's basically the perfect system for me

 

Right, you've confirmed what has been noted all long. Thank you for that much.

And I will tell you that simply is not true. It does not have to be sustained, when the CPU breaks the 100% threshold on all 4 cores, the GPU immediately throttles and then goes back to full speed and throttles again. This is evidenced by everything on the first page and subsequent posts.

Again not true. The GPU when pushed to it's maximum using furmark, never exceeds 75-78C. Asus G51, which has MUCH less optimized cooling does not buckle under any thermal stress and its temperatures can reach up to 100C. Remember, the thermal maximum for an nVidia chip is approximately 110C, the M15x never comes close to reaching that peak under ANY kind of stress. The CPU TJ max is 100C and it never comes close to reaching that temperature under any sustained load either. The thermal excuse falls flat on its face.

According to you perhaps. Others that want the most of their system would disagree. If Dell is marketing the Alienware M15x as a premiere gaming system, it should perform as such. Running a CPU at 100% load while far below TJ Max should not hinder the system in any way. Furthermore, competing gaming systems with similar cooling designs (I'm looking at you Clevo) can sustain peak 100% CPU/GPU loads all day long. I guess what you are saying is, "if you guys want a real gaming system, buy a clevo".

That statement contradicts itself. It eats games for breakfast except when pushed to 100% usage, at which time it buckles. That makes it a failure for which it was designed. My desktop with a C2D and 8800 GTX can run at 100% load all day long without breaking a sweat. My previous G51-X2A could also do the same and that was a budget gaming system with an inadequate cooling system. What will Dell do when the M15x is pushed to 100% peak load with a game and/or application? Will it refund anyone that hits such a wall or will they be advised that the M15x was not designed to do so and to upgrade?

Edit: Why not do this: Create a beta bios program, have power users test it out on their M15x with the throttling limit removed and have them sign an NDA. If no damage is sustained, then you could release it to the public. If it is, then you can release a statement saying you have tested the issue in the lab and with end users and have found the system cannot sustain 100% loads. That is what many other companies do so I cannot fathom why Dell would not. Community outreach should be your first method of ever attacking a problem such as this.

 

Also from my experience, a LOT of programs will max out one or more CPUs, even when it doesn't make sense. Power DVD 9 completely maxes out one core. Many games completely max out one or more cores-I see that all the time. Gratned, this apparently has to max out 4 cores, but that WILL happen at some point even for programs that aren't really taking full advantage of the cores.

 

Exactly and the implication of this is simple: you will have to pick and choose between multitasking or using the M15x for a single application. The i7 was designed for multitasking, hence the 4 cores, but the artificial limit placed by Dell prevents the chip from ever being utilized for which it was designed.

 

While I appreciate that a Dell representative took time to discuss this with us, I am extremely upset at the position taken. I don't give a rat's what your opinion is, why you have them, or how your laptop performs with your applications.

I do care that my ALL POWERFUL M15X has the potential to choke when stressed in a way that, while uncommon, is not beyond the day-to-day uses for the people it is marketed towards.