AW M15x Throttling Issue Investigation - Stock clocks and overclocked.

@LeWZ
Hey, here are the power consumptions for the *popular* 3 cards..
GTX 260M: 75 watt
Mobility 5870: 50 watt
Mobility 5850: 39 watt

There are several people on this board has replaced their 260Ms with 5850s and 5870s and I haven't seen any issue being reported by them. However from what I understand 5870 gets a little hot sometimes but fans handle it. I don't have any experience with such manual replacement so I can't comment much on this one :/

 

That is a interesting point... On the Dell site the driver for the 260M says 65w
Also I have noticed more problems when using the nvidia drivers from the nvidia site.. So I would suggest using the Dell drivers for now.

I am kind of interested in testing a ATI card now LoL

 

Nothing better than a "friendly" discussion dude
I prefer nVidia too but I don't know.. they are just messing up since 2009 and I keep hearing new 2xx.xx series drivers are of poor quality for games(note that I haven't tested 2xx.xx myself). I don't even want to remember the 19x.xx series problems with Fallout 3 :/ My belief is that nVidia is moving its focus from "cards for gamers" to "cards for bussiness" as you can see by the increase of R&D money being spent and advertisements by nVidia on CUDA, Tesla and several other mostly computing related areas.

On a side note, I don't know the actual power usage of 260M in M15X, I simply copy pasted the values shown there:
Notebookcheck: NVIDIA GeForce GTX 260M

BTW I was interested in FSU a while ago. But I thought its over-advertised, what it is like? I don't know what to ask to be honest but do you get to work on real stuff? research and development or more "industrial"?

 

What we really need here is someone with both cards to do a real-world, multi-game test. In game benchmarks, and out of game benchmarks, from the same system and same hardware. I am getting my 5850 tomorrow, and if return service is not demanded for my 260m, I will gladly run the tests.

 

awesome that would be a great comparison so that we can something to base upon a same system test unlike separate systems being compared my not have the same internal hardware like a new mobo rev to a newer etc.

 

I spoke with a tier 2 specialist @ Alienware and they told me that for the previous laptops sent out there are no "official fixes" for the throttling. For all laptops made within the last 2 months, they say, do not have the issue.

Also - NVidia vs ATI: Nvidia generally has much more stable drivers. ATI may have some 1ups on the cards, but with better driver support it actually ends up in NVidia's favor on in game performance (according to the alienware t2 tech).

 

This is an interesting statement, because before Dell came out and replaced my CPU and motherboard, about a week ago (My laptop is 4months old), the throttling indeed was much worse. The overall performance of the system went way up with the new mainboard and CPU, however, throttling still occured, just not as much. So maybe there is something different with the Rev. 11 + motherboards compared to the older ones.

 

Just an FYI: Everything I ever read about the Area-51 M15x (first edition) was terrible. I'm currently sporting one of those with an 8800 GTX and after 3 years it still works perfectly. I've never once had a single problem with it. I ordered my new m15x because I'm due for an upgrade.

 

So who knows, their statement could be true about the lack of throttling since Rev. 11. Maybe I was still experiencing throttling because they swapped the CPU, mainboard, but not the graphics card.... We will just have to have someone that recently got theirs in with the 260 run GPU-z and check it out.

 

That is not true... I just got one a month ago.. Brand new.. still throttles.
They are working on a fix but it is in testing.

 

thenzfarmer... Have you tested that with Prime and Furmark running at the same time?

 

Give me a few hours

 

Not just Dell either-reviews totally ignored problems with Asus' G51 series too. Which is why I don't pay a whole lot of attention to professional reviews...they rarely have useful info, like say...THE NOTEBOOK IS FUNDAMENTALLY CRIPPLED/BROKEN :lol:

Thank goodness the M17x is apparently fine.

It's inexplicable to me how Dell, Asus, etc. can have a major issue with a model and not address it.

 

sweet await your test as for me so far (laptop 1 week old) its not performing well. i have the 5850 i7 740 and i am sure something is throtling the gpu

http://forum.notebookreview.com/ali...-1-m15x-scores-260-vs-5850-a.html#post6555986

thats the results i got, will be very interested to see if you get the same or hopefully much higher so i can see what to aim for. atm dont see any bech marks for 5850 on the m15x

 

So even the 5850 is throttling?

 

First off I must tell everyone testing throttling or gaming with a 260M etc.. DO NOT USE NVIDIA DRIVERS FROM NVIDIA's site. You can if you want however after extensive testing shows that most other drivers then the ones supplied from the Dell site make the issue worse.. So with that said some might reduce their throttling greatly simply by using Dell drivers. I assume many didn't even suspect this as normally this is not an issue however I can confirm this!
(ya ya I know everyone wants the latest drivers, but if you have a problem try the Dell ones if you are not currently running them)


2nd thing I would mention.. If you want to properly test throttling like with the 5850..
Use Furmark with the suggested settings on this thread AND run Prime 95 at the SAME time. If you are just running Furmark you are not testing the CPUs fully. If it marinates with both running for a while with no issues then you are 100% throttle free.

Try PRIME at 8 threads and 6...

Both of these things seemed to be commonly overlooked while testing.
If you have the wrong drivers Furmark alone easily throttles with some configs. If you have the right drivers or have an ATI and just run Furmark alone you might not be able to see the throttling either or it happens at lower intervals.

In gaming since it is much more dynamic then synth tests and it is easy to write it off and say "my system isnt throttling"

Anyone doing these tests please post results or screenies.. All of this helps me!

And the testing continues...
Thx

 

I have spoken to someone that indeed has been able to make the systems with the ATI throttle. However in my opinion some new data would be nice so... please read above if you have a 5850. I might just try to get one sent to me for giggles and testing.

thx

 

Sorry guys my BAD, the 5850 isn't throttling.

Read my updated post for new scores.

http://forum.notebookreview.com/ali...-1-m15x-scores-260-vs-5850-a.html#post6556288

there is a setting for ati called " power play" which you can enable to throttle the ati cards down in certain conditions.
however if you turn it off it ALSO throttles down by default to about 50% of both memory and gpu speeds and fixes at that.

So upshot is leave it on.

 

His issue was caused due to PowerPlay settings. It seems like he corrected the settings now.
http://forum.notebookreview.com/ali...-1-m15x-scores-260-vs-5850-a.html#post6556288

 

Hey, coming back to testing with the Watt reader.
155 Watts stable if running Furmark and Prime 95 simultaneously (went as high as 157).
135-145 Watts if running Prime 95 only and no 3D application (apart from windows aero).
128 Watts if running Furmark only and no CPU hungry applications in the background.
65 Watts at Idle.

All these with RivaTuner set to maintain stock GPU clocks (550-1350-950).

 

Have u ran furmark with the settings listed on page one of this thread and then while furmark is running start prime95 with the settings suggested in this thread?

 

erm.. isnt that not good? I heard that if the power draw is higher then the adapter then the battery compensates.. but still that seems like its not good.

Also did you have many USB devices plugged in? plug em in!

 

here is furmark+prime with 5870

 

Run it longer please 2 mins isnt anything. I would suggest 10 - 15 would be golden ;p

But good thing u posted this..

 

hahhaha here is longer one. looks like cpu is throttling. if you notice the chip% its at 37.5%. I'm sure if I enabled throttle stop then it would be fine.
will post a 260m test later.

 

Clock modulation and chipset clock modulation, CMod% and Chip% on the right side of ThrottleStop, should never be less than 100% unless your heatsink has fallen off of your CPU.

inap: Both of your log files show massive amounts of throttling. I'm not sure what Dell has been up to since this model was released but there is obviously still some significant problems that have not been addressed. Even at idle, chipset clock modulation is telling the CPU to slow down and throttle.

That just doesn't make any sense. How is your testing going DeeX?

 

Also, what rev. # mobo are you running, which driver are you using, and which bios version are you on. I noticed a massive increase in performance, and reduction in throttling on the dell drivers, A06 bios, and I'm on a Rev. 11 mobo.

 

A 24 multiplier looks great but when you multiply that by 37.5% to compensate for chipset clock modulation, you get the equivalent of a 9 multiplier. That's not too good.

 

yeah i was real surprise to see it throttle that much,but ever since A06 throttling has become hugh for me for cpu and gpu. i can't use the 260 at all because it throttles so much.

rev.11 mobo, bios A06, driver is 10.3

 

Many have that Rev. Nothing rly changed tbh...

Test with BIOS A03.. A05 and higher actually made the throttling worse and more sensitive.

 

I think this speaks for itself. While Running BBC2

 

yes it does

 

I'm going to cancel my M15x and just get the M17x. Not going to deal with this throttling bull. Its ridiculous that Alienware would even sell a non functional computer like that.

 

Wow, so the 5870 throttles as well. Probably the 5850 will likely throttle too. No escape for m15x owners then? I was planning to buy a 5850 or a 5870 to replace my 260m and solve this issue for good. But I'm rethinking this option now...

 

5850 isn't throttling who is making this up?
And as for 5870, it isn't part of the standart system setup so you can expect problems with it.

 

Do you have a 5850? If you do, could you run furmark + prime95 while running gpuz and post your results?

 

look two posts above ur post lol.

The 5850 Throttles too.. They both have the same power consumption.. Matter of fact the 5870 is just a overclocked 5850.
The only difference is the ATI cards only throttle the CPU, so you cant see it with GPUz you need to use the logging in Throttlestop to get a idea.

ITS NOT THE VIDEO CARD PEOPLE!

The M17x isnt a perfect machine either, I would poke around and talk to some users. I was considering that as well this week and already two users explained to me some issues.

 

The clevo card doesnt throttle, because the BIOS doesnt know how to throttle it. I cant vouch for Dell 5850 cards. It seems though that the 5850 throttles less (if at all). Also note, Slone has overclocked his card. It might be that on default clock it doesnt throttle as the power consumption is smaller.

 

Stock 5850 isn't throttling. As Wattos said Slone's card is overclocked as you can see "750mhz ??".

As you said it is not the video card but your above sentence makes it sound like ATI causes the CPU to throttle Its totally CPU and BIOS's decission to throttle lol. They got choices and they choice to ruin your game experience. But with ThrottleStop, they don't have a choice.

 

I just meant that often people are thinking the problem is some how related to the video card. One clear reason it could be is the lower power consumption of the ATI cards. That doesnt mean that they wont throttle. If you can get it to synth throttle then it may in the future throttle on some new game or whatever.
If there is a defect in the m15x it self and you just use a lower power consuming card there is still a chance that card will throttle too as the problem hasnt been fixed.
I personally think its a hardware power issue related not the bios decision. I think if there is a bios fix that it is just a bandaid to avoid having to do a hardware revision / new adaptor. I hope I am wrong though....

Well then heres and idea... run a 5850 with Prime95 on Small FFTs 8 threads + Furmark on Xtreme Burn and post process at the same time, and with throttlestop logging this and turned off of course.
Run this for at least 15 mins.

& then post your throttlestop log please.

That will end this debate for sure. Someone plz Doooo it!

 

Then I'll do it tomorrow

Yep its related to ATI's lower power consumption. Every card throttles however main difference is at what level it starts to throttle.
Comparing 260m and 5850 isn't fair because 260m is 1 year older than 5850. Therefore 5850 will start to throttle in games in more distant future compared to 260m. However it isn't logical to call this as "throttling" as it will be caused by GPU's inability to satisfy new games' demands.(As throttling refers to GPU itself downclocking for a while)

I think most people is missing one big point about Furmark. Furmark -unlike other benchmarking tools like 3Ds- is a fatal stress test, it is loading your system maybe atleast five times more than what games will be able to do 4 years later. And I'm not planning to use the same card 4 years later lol. Still I'll run Furmark and Prime95 probably tomorrow, then post the logs here.

As a side note to other folks, whatever you have an ATI or nVidia, your CPU can throttle and it isn't directly related to your GPU.

 

Well played! I think there is much confusion and a lot variables going on here. Plus some fanboyism at play.

Again I must stress to anyone doing proper testing... use official Dell drivers not ones from nvidia. I am not sure about ATI, however what I am sure is laptops are not like desktops and even the cards sometimes have proprietary video bios(s). Because of all that they have drivers spacific to the systems.
Dont believe it? Well a Dell engineer told me for one, plus my testing shows this as well, and lastly the link below touches on the issue:

Dell Certified Drivers

For some of you simply using Dell drivers can help reducethe thottling, some of tests I have done show that when compared to using the latest nvidia drivers.

 

here is same test with 260m. dell driver, and bios A06. once the test started the 260 would downclock and rarely upclock back.

 

wow a clear case of throttling

 

but look at the awesome temps!

 

hahaha just trying to make sure high temps is not the cause for the throttling.

 

your ATI card, is it a clevo?

 

the last test is with the 260m the previous test was with the clevo.

 

Note, the clevo card will never throttle as the bios doesnt seem to know how to throttle it. If you press the stealth button or run on battery, the clock will stay the same (as reported by GPU-Z).

Therefore we need somebody with a Dell-ATI 5850 to do the tests.

 

yeah i know but deex wanted me to test for him so i did it.