AW M15x Throttling Issue Investigation - Stock clocks and overclocked.

Info no longer relevant. Dell has released a new bios update that resolves it.

 

you can run OCCT to have it print out in a big chart and the temps will fluctuate along with the speed...theoretically

 

is this a big issue, man just ordered M15x and now I hear this. hopefully not a mainboard issue.

 

Until others confirm this is happening to their systems, it's not definitive. Unfortunately not a single person has stepped up yet.

 

I will test mine..when it arrives! :complains: Which CPU do you have again?

I'll also post some tests using my Kill A Watt power measuring tool. It would give us a better idea of how loaded the AC Adapter really is.

 

Agreed, run the OCCT PSU test and graph it... see what you get.

 

I ran OCCT PSU but it only shows temperature changes, not frequency throttling. I did however run GPU-Z at the same time and this is what it looked like:




The spikes are when it hit full stock clocks and the rest was the throttled 3D clocks.

 

is the M15x power brick 120W (19V/6.3A)? if it's like the G51J, then AW cut corners on power. the graphics chip requires 75, CPU @ 55, and extra for peripherals (screen etc.) which go well over the 120W with my G51.

log CPU freq/multi with http://www.cpuid.com/tmonitor.php while having those running. my guess is it will be throttled when the GPU is used and vice versa.

> http://forum.notebookreview.com/showthread.php?t=436536 says 120W

 

No, the M15x uses a 150W PSU.

It probably requires closer to 180W.

 

Hey thalinx, the PSU is rated at 150W. I believe the old area m15x used the 120W adapter. I'm going to do a respawn (factory restore) to rule out any possible software glitch and then run the Tmonitor with the benchmark programs. I'll update this thread once I do.

 

Your 210W comes tomorrow right? I'll be waiting for good news (And demanding Dell give me a free one once this is clearly the issue).

 

150W is stretching it very thin. if it's a lower quality PSU, then at usage that high it might fluctuate too much to be stable (hence the underclocking). if any case, i would expect Tmonitor to show fluctuating CPU clocks as well, in sync with the GPU.

i wish asus would give a bigger brick

 

I have the same brick as you. I don't experience any throttling however, as my T9900 uses much less power.

 

Monday is what it says. They didn't process my first order (which would have arrived tomorrow) so I had to put in a 2nd one. BTW to add to the bad news, I underclocked the GPU to 450/850/1125 and it still throttled. I don't know what to make of that, either it's a power issue or something more nefarious.

 

Well this an Asus flaming into IBT+Furmark Extreme burning, 8xAA.


This is my Killwatt meter, screen while doing the test


It spike to 138W, while mantining between 119 -135.

 

what if you try with the clocks only slightly over 3d low, 385/300/770 +50each.

 

My G60 does not throttle either. Heavy overclock on the GPU and 3.3ghz on the CPU. 32 threads on INtel Burn Text, 1920x1200 16x MSAA on Furmark.

Chipset starts to toast at 90C and I'm done. GPU seems to top at 90C as well.

 

Are you serious? My M15x better not be here before your PSU gets here... I need to know the "truth" before I open that box...

 

This is pretty bad, the flaming Asus laptops are holding up fine but this uber M15x can't even hold a steady clock under load. I just did a complete system restore with everything stock and tested with P95 + furmark and it was the same throttling. Definitely not good news and that pretty much rules out any software conflict that could have arisen from me installing programs and/or different drivers.

I REALLY hope the 210W PSU solves the problem but I have my doubts at this point. It seems it will either take a bios update to remedy it or even a motherboard revision. It would be nice if other M15x users stepped up and did some testing. I see so many people complaining about ridiculous things like stuttering sound or bad customer service yet they don't seem concerned enough to test their systems for throttling--which completely defeats the purpose of this laptop!

 

Honestly from my standpoint it sounds like a bios issue.

 

The only good news is that I don't think many real world scenarios will arise that causes throttling... (even though my G60 does not throttle with everything thrown at it, I would never run something that puts my Chipset to 90C, my computer would be dead in short time).

Here is one more test that might be helpful:

3DMark06
3DmarkVantage
Crysis Benchmark (or similar).

Crank every setting you can and log the GPU-Z screen shots and see what happens.

 

Hopefully it is as simple as a bios fix. I don't think anyone should have to throw down another $120 to gain full potential of their system.

 

Well Joker I'll test mine as soon as it arrives tomorrow. Until then I'll continue posting about sound glitches and crappy customer service

 

Apparently there is a Dell engineer/rep on the forums. I PM'd him and asked that he look at this thread and notify the Dell/AW engineers of this potential problem. If we can get more M15x users aboard to test, this will help get the issue resolved quickly since Dell is already actively looking into the Inspiron/XPS 16 throttling problems.

 

the only issue i have run into with mine is the BSOD's with my esata drive, but i think that is either the drive controller or e sata cable.

 

Have you run the tests I suggested in the first post?

 

Yeah, so far so good right now, I will test some more for you tomorrow bud!

 

So your M15x does not throttle?

 

Make sure to run the tests exactly as outlined in the first post. Screenshots, system configs and drivers used would also serve as vital info. Hopefully this is an isolated issue.

 

I'm surprised Prime was able to throttle your system but not Burn Test with 8 threads...

 

Yep I found that strange too since burn test is supposed to be more stressful on the CPU.

 

Just read the last 15 pages on the XPS 1645 issue that you made a post on, and it has a lot of good information. Assuming it is the same issue it looks like it may be a bios issue. Just ordered the system myself so I can't really do any testing, but any chance anyone has one of those Kill-A Watt things to test how much power is being drawn?

 

Just because of what you said, I have my contingency plan....

1) keep my desktop PC with core i7 920 and 6 gb ram ATI 5870

2) ALso have my Alienware M15x

3) Problem solved

 

well that is the point of helping...to educate other ppl on stuff they don't know. and how am i trolling -_-, i asked u a question, based on what Panda has said previously...

and now my guess is, run GTAIV and see if u throttle since that is a quadcore capable game, rite?

 

I don't own the game but hopefully someone with an M15x that does can test. I just finished installing L4D2 so I'm going to crank it up to 900p max settings and keep GPU-Z's log running to see if it fluctuates. Same thing with MW2.

 

Reading through the SXPS 16 thread, it looks like the system is limiting power draw to 90W. I can't imagine this being the case with the M15x though, or they wouldn't come with 150W Power supplies.

If it isn't a system limitation, then the 210W power supply should solve the problem. I guess we'll know definitively on Monday eh? ;/

 

Or until those cowards in the M15x owners lounge decide to be a man and try to reproduce the issue with a PSU test.

 

Yup either until Monday (provided Dell ships it on time) or other M15x users test their systems to determine if it's just isolated to mine or a bigger problem.

 

Im not sure compeltly, but i read there was an issue with IBT n i7, i think was with the desktops, but yea, since early i7m i saw that even 3dmark took it above, still on c2d ibt goes higher than prime95.

 

Ive been following this thread for a few days now, from the M15x lounge, and now here. I have yet to notice throttling, that I know of. I've been running Dawn of War II graphics test and at native 1600x900 get peak of 86 fps and drops to 12 fps under intense battles, but, it only hits 12fps for asecond or so. AVG is 35fps with all settings nearly maxed, minus AA and shadows. I will load up vantage and '06 tonight when I get off of work and run it through the ringer. I will only use stock clocks however, i don't OC anything, no need to really. I'll post results when Im done or early tomorrow morning.

 

Thanks! Also if you could run through the synthetic tests as outlined in the first post that would be very helpful.

 

Will do, could you provide links, or are they on the first page?

 

Since my laptop is arriving tomorrow and I don't want to raid in WoW could you clarify something for me? Do I just run the Intel burn test (32 threads, maximum stress level) and Furmark (1600x900, 8xAA, stability test, post processing and xtreme burn) at the same time? What specifically am I looking for with GPU-Z? Never done anything like this but I'm more than willing to help. Surprised no one else has yet.

 

The instructions are all on the first page. I'll update the first post to include links to the programs.

 

I just did a run with TMonitor/GPUz active for logging, Prime 95 + Furmark. Attached are the log files for the run, take a look at the GPU-Z one, I noticed that shortly after it throttles, it lists the GPU load as >100%. Interesting huh?

 

Interesting indeed. My question is this, Could this have anything to do with the stealth function, or any of the Alienware proprietary software. I'm currently running my unit naked (Win7 upgrade), no alien software on it. I'll see if that has anything to do with it. You might have already tried this, but I understood your previous post as that you did a factory image restore. I wonder....Thinking about this whole thing, could it be something to do with the hyperthreading of the I7, maybe what we are seeing is the cores switching on and/or off causing a fluxation in the power delivery....I've got some things to try tonight, won't be until 0200 in AM unfortunately.

 

Just did a preliminary test. Loaded up GPU-Z. Let the sensors run for a bit, then loaded up Prime95 (Max heat/ Max power) and the clocks immediately throttled back. This is with a stock Alienware install WoW edition.

Update: Strange that GPU-Z didn't acknowledge the adjusted readings after exiting Prime95. The sensors stayed low, but exiting and relaunching GPUZ showed the 260M clocks were back to normal.

Update #2: The downclocks vary depending on how fast the fan is spinning too. As the heat increase and the fans increase, the downclocks appears to me more severe.

Update #3: Ok, letting Prime95 really settle into "Yes my fans are definitely working" max speed. If I exit and re-enter GPU-Z, my clocks speeds are at max and appear fine but eventually downclock, but again, if I exit and re-enter, clock speeds are fine for quite awhile. I'll try some of the other tests, but right now, GPU-Z is suspect for its readings, as exiting and restarting the app will show the clocks back at normal for a variable amount of time.

Try running your tests and exit and restart GPU-Z numerous times and check the sensor readings.

I do think clock speeds are reacting to the power load of the CPU/system and it is walking a fine line. It will be interesting to see what the 210 adapter provides.

 

Here is a pic of GPU-Z with WoW running sitting in Dalaran (graphically and CPU intensive on a single core). Notice the dips and spikes. Yet frame rates remained consistent (54-62fps) and did not correlate to the big dips and valleys.

I might try a clean install to see what I get versus the Alienware factory install.

Doing raids, performance is very consistent and even (well as even as WoW can be, heh).

 

Electrosoft, if you're up to it, try overclocking your system and see if the performance scales with the increase in clock speed in 3dmark with GPUz open. I noticed that my GPU performance did not increase passed 610-620 core.

 

Just did a simple test with DoW II graphics test, everything maxed, no AA, with GPU-z logging, and there was indeed throttling.

However, I timed the GPU-Z with the game launch, menu option, selecting test.... noting via stopwatch when said events occured.

At launch the card remained at 200 clock
During the intro movie, it peaked at 550 clock
Once at the option menu, it dropped to 200 clock
During the actual demo, it reached 550 clock and remained there throughout. The fans reached speed, temps reached around 62C.

Upon quitting the game, the clocks returned to 200 and temps reached a balmy 51C within 3 mins.

I know this isn't as conclusive as the Prime95/Furmark testing, but this game does stress CPU/GPU to an extent. If anything, this shows that there might not be any clocking issues during realworld gaming.