[Conference Call] Nvidia GeForce GTX 580M 78C° Throttling [SEE POST #191 and later]

@Widezu, glad to meet another Rugby supporter! I actually took my first real vacation in 8 years to go to the World Cup last year and we had a blast...GO ALL BLACKS!

@Micky I hope it doesn't come to breaking out an oscilloscope here as I really don't want to get that far into this. I just think with a few more bits of information we can figure this out one way or another. That's what I'm hoping for anyway.

FYI, I did a bit more testing on battery power this afternoon and depending on what info Dell can provide us with regarding the voltage regulation I am becoming a bit more concerned.

My system on battery with the GPU at P0 50.6/67.5/101 is drawing roughly 40 watts @ 12.1 volts which of couse is completely throttled down while on battery. During this period the CPU is definitely "Racing to Idle" jumping up from core to core as shown by HWINFO. For those of you wondering what I am getting at here google "Dynamic Frequency Scaling" as that may help, Wikipedia has a good little article on it.

Based on the idle state, battery capacitance and time elapsed to drain the battery (I won't bore you with the math) I figure at full load the system would need approximately 218 watts at 12.1 volts (.8 volt on the GPU @ idle and .87 @ load). In other words the battery would last for roughly 40 minutes off a full charge.

Now I see why Optimus is so important as without the IGPU the battery drain is very high even at the lower P state. Essentially, given my measurements, the battery life (with the IGPU disabled, Optimus off and the 580m at idle) is about 2.25 hours, not that it does any good having the 580m on at all unless you're plugged in to the wall. The battery life doesn't really surprise me as this is a DTR but what I am concerned about is what power requirements are actually necessary at loads over 85% and whether the power supply and the system itself can meet the individual component's demands without voltage drops and/or excess heat generaton.

Now the catch here is I am working backwards from the battery capacity, observed system voltages and observed drain down time so unless Dell is willing to tell us the design capacitance of the individual system components we can't (as end users) really figure much of anything concretely but I am concerned, like I said, that the "horsepower isn't there to pull the wagon."

Tomorrow is Monday so hopefully we'll get some more info during business hours.

 

Well your calculations on battery life are spot on (for me). The 3D model has the IGPU disabled (not wired in) so no optimus and I checked out my battery life in max power save setting and see exactly 2.25 hours just sitting in windows with the GPU in P8...

 

Dell always denies at first. And then when they are caught with their pants down they would lie to conform to their fault. At the end its about how much they are willing to spend to rectify the root cause. Dell always tries to give known problem for which they refuse to rectify, a colour of high technical specificity and secrecy in the name of business.
My take is that Dell doesnt want to rectify the problem on the existing machines as well as those which are already in stock. The cryptic lingo is absolutely known Dell "modus operandi". They would drag this long enough until they release R4, leaving the R3 owners high and dry. Dell has repeatedly done this in all their lines when facing with chronic issues.

 

I think you're dead wrong when you say "always," especially in the context of this topic. Feel free to read the thread and find out where we have explicitly denied the GPU doesn't throttle at 78C, which is the topic of the thread. Also, coming into this thread and asserting Dell is lying about something I take as an insult.

As far as the root cause of 78C throttle, no amount of money Dell can spend will change it. We have a spec. They know our spec. One company hits it. One doesn't.

If you're referring to cryptic lingo from Dell in general, I can't comment without context. If you are referring to my cryptic lingo, I've said about a dozen times there's a reason for it. Feel free to search this thread for all my posts before you accuse us, within the context of this thread where you posted your assertion, of not being open, supportive and forthcoming. Also, feel free to find another PC vendor who will put the lead GPU engineer on the phone with their customers for an hour and a half- and no NDA, by the way.

 

Bill has been trying very hard to help us out. We've got a conference call set up soon so Dell can talk with us, discuss options and give us better explanations (in case you haven't read this thread). So, having wasted your first post here on this magnificent forum to flame Dell and Bill, I tell you politely GO TROLL SOMEWHERE ELSE

 

^^Well said

 

I can't think of a single company that would bother arranging to have their low level technicians available for a conference call let alone a lead designer. You guys have been awesome, I plan to jump in during the next conference call and listen in and emphasize to Louis that we need these throttle limitations removed in future designs. I know most designs are at least 1-2 years ahead so I'll hold out hope for the redesigned models (2013?).

Actually its quite the opposite. Over the last 2.5 years that I've been on NBR, Dell always had someone available to engage the customers and address their concerns. I mentioned in a previous post that we had to fight to have issues resolved in the past but they key word here is resolved. They didn't leave us hanging and after working with the customers, they fixed the problems. When the M17x-R2 was first released, it suffered from an oversight in design that caused it to prematurely trip its throttle point. We brought it up with AW and at the time, the AW guy in charge invited me to their AW headquarters and had one of their engineers personally come and capture the system. When I went to pick it up, they gave me a tour of the facilities along with some free AW oriented goodies--what other company do you know goes out of their way to do this? Try asking Apple for a tour of their Cupertino HQ if your MBP breaks down and see the reaction you get. I'm confident they'll sort this out and be glad they want our input.

As customers we definitely have a right to feel frustrated, I certainly am with the throttling and with Bill arranging a conference call with Louis, we can convey our concerns to him and hopefully he'll be able to do something about it.

 

^^^ As always, eloquent and to the point. Props and +Rep.

 

I will not say that Dell is innocent of any wrongdoing when it comes to some issues. Even while I worked for them, there were times I could see things that Dell wanted me to do or say to customers that made me upset as a tech.

To understand the "modus operandi" of Dell, you cannot think like a consumer or like a tech. I see and understand more of that now that I have been through many college courses for becoming an IT Manager/Executive. Yes, every organization has kinks. Hell, every corporation has people in middle management that even the C level brass knows does not belong there. The point is, Dell walks that very fine line of being a business and in it to make a profit and caring about their customers.

Dell doesn't deny first, they observe first. Yes, sometimes to a fault but you cannot say that everyone on this forum is a representative sample of even 5% of all Alienware owners.

Dell has amazing engineers, designers, and a lot of the L2's are damn fine techs. Some of the phone support guys are pretty good too. No other OEM would be as responsive to an outside community as Dell has been with the people in this forum. HP would laugh at you and tell you to suck it up, if they responded to you at all.

 

Since my PM isn't working yet can anyone tell me if the conference call is still planned for 3/22 in the afternoon?

Thanks in advance!

 

Yes, it is, you have over 5 posts now so I think you can PM if you want in the call.

Also, let this be a last call to everyone, I will probably be submitting the list tomorrow. If you want in, or even think you want in, PM me your name and email address. Without an invite you cannot join.

 

I dont think I have enough posts for pm yet lol, but I want in. Would like to hear whats going on.

http://forum.notebookreview.com/gaming-software-graphics-cards/261929-disable-powermizer-tested.html


this should be integrated into the discussion here. I have a basic, possibly older, version of this running on my system I current use an m17x with a 580m that throttles badly, and fluctuates quickly a few degrees every few minutes. Already tries a TON TON TON of things, and while this doesn't fix the issue, it definitely helps.

 

Dear Bill,

First, thanks for all your personal efforts.

Nonetheless, I hope for Dell's own sake that this issue gets resolved quickly. After reading through this thread and related ones, I seriously started to doubt the M17's build quality and almost cancelled my order (also because of the unexplained 580m delivery problems).

The only thing from stopping me has been the manner in which you and much more knowledgeable forum members than myself (big thnx for your efforts guys, much appreciated!) have picked up on this issue.

Good luck on the conf call all!

Waiting for: M17X R3 Stealth Black / 2760QM / 580 M 2GB / full HD 120 Hz screen / 256 GB Dell SSD / 8 GB 1600 php / Blu-ray

 

@F-117 You all are most welcome. Alienware is not one of my team's platforms, but I love the hardware so much I jumped into this thread. Going forward, Chris and Jesse will be watching and working the AW boards. They each have 12+ years at Dell, most of that working on our forums. So you're in good hands. You made the right call sticking with your order. I'm pretty sure we can all sort this out and come up with the best possible solution, considering that mysterious Dell spec I keep referencing, which we will talk about on the call.

 

Again, many thanks Bill!

On a different note, can either you or Chris or Jesse please shed some light in the "Order status discussion" thread on the reason why it is so hard for Dell to get the 580m back in stock again. I believe many people are looking for an answer to this question.

Best, F-117

Waiting (for 4 weeks) for: M17X R3 Stealth Black / 2760QM / 580 M 2GB / full HD 120 Hz screen / 256 GB Dell SSD / 8 GB 1600 php / Blu-ray

 

Just make one more post anywhere and PM me your info by the end of today.

I will probably make the end of today the cut-off, I will be submitting the list tonight before 8PM MST.

 

I can absolutely check on 580 service stock and new build supply issues. I got a guy.

 

I know the 580s are running low. Just goes to show how much demand there is for them am getting mine replaced in an effort to fix my problem. I already had my heat sink and gpu fan replaced, but it only maybe helped a degree or two. So now the tech I was working with was able to find me a replacement.

I gotta say that the techs and many other people I have been working with have shown that while there will always be tech problems , dell and AW have people who actually care to get the job done and take care of the customer..

 

Ok, the final list has been submitted, everyone who PM'ed me before now is on the list.

The final details will be sent out to your emails, which will include the exact time of the call as well as instructions on how to join.

 

Hey Vytis your posts were some interesting reads...over my head but still interesting. I was wondering about your repaste and repad job. When you tested your new temps did you elevate the back of your laptop at all? i keep mine elevated on stock and get 60c at full bf3 and crysis 2 load. sadly when i remove the elevation it raises to the 78c throttle. im definitely considering a repaste. second question: what type of thermal pad did you get? who makes it and how thick is it? thanks

 

Thnx Bill, I will inform the guys in the "Order status discussion" thread of your answer.

Waiting (for 4 weeks) for: M17X R3 Stealth Black / 2760QM / 580 M 2GB / full HD 120 Hz screen / 256 GB Dell SSD / 8 GB 1600 php / Blu-ray

 

Louis has a kiddo out sick this week. I'm waiting to make sure he can still make it. I'm assuming he will be able to make the call from home, but waiting to get the official thumbs up before I send out the invites. Should be out this afternoon.

Edit:
Oh, and I think Louis can talk pads and paste on the call as well. He has some good info to pass along.

Edit 2:
Louis is good for tomorrow. I'll send out invites with dial in information this afternoon.

 

Just had my 580m replaced (1st time) and my heat sink and fa replaced again (2nd time). I am still getting fluctuations on my gpu temp. Going up and down in a roughly 5 degree range. Literally just goes up and down and up and down and up and down , all around the median 55C temp. Then I switched on the custom fan setup through HWiNFO64 and now I am in the 40s and seemingly leveling out.

I am going to do some testing, and put the system through the paces, but Im thinking now its just more of a fan issue than anything else, whereas before, even with custom fan setup I was running 55-60, now I am at 43 idle.


So my question is: Would you guys be so helpful as to check your idling for me. What do you sit at when just browsing the web or at the desktop? ... and how much , in say a five to ten minutes period of time, does your gpu temp fluctuate (assuming no custom fan setup or other mods like that)?

Any helpful info would be greatly appreciated. My main concern is that the big fluctuations may be due to something with the motherboard thats not controlling fan speed properly, since with controlled fan speed it, so far, seems to be ok. Leveling out now, it seems, at around 42C.

 

@Phoenix that is completely normal. You are looking way too into this. My GPU ranges between 50-60 when doing everyday tasks. It is the powersaving feature of your GPU kicking in ramping it up when it needs to.

 

Ok. Just wanting to check, since I dont have a baseline to compare it to.

Since I got the 580m changed out and been running the custom fan speeds, I have dropped into the idle temp of around 38C. Whereas without custom fan speeds I am around 55C.

Have not tested in games yet, but am about to. Glad to see such a drop, since beforehand, even with custom fan speeds, I was sitting around 55-60.

 

It sounds like with your new card you probably don't need the custom profile, except perhaps in ultramodern gaming. You should try using your laptop without it.

The standard EC doesn't even kick the fan on until the mid 50's, and doesn't get serious until the 60's. I would say a normal max temp for most games using the standard profile would be 65C, with ultramodern DX11 games pushing the temp higher.

 

I still get around 70 with the custom fan setting, and just would rather keep it coolest I can. Will probably try it gaming without the custom fan setup and see how it goes.

 

Speaking of Dell and thermal profiles... XPS 13 BIOS updated.

Ironically, the fans were a little too aggressive...

 

Did anyone get their invite to the call?

 

Sending out now. Sorry. I got pulled into another project this afternoon. @brianwi I have yours coming as well.

Edit:
Invites out. We're looking forward to it.

 

Hopefully everyone got their invites, make sure you guys understand what time the call is taking place. Unless I am mistaken in will be at 2PM today Mountain Daylight Time. Here is a link that shows the current MDT, so you can figure out when that is for you locally: MDT – Mountain Daylight Time

 

good luck

 

Looks like MDT is currently GMT -6 which is also the standard time for the Central Time zone. So will be 1400 @ GMT -6. Unless I am wrong and getting MDT and MST confused...ugh, the one thing I hate is daylight savings time and how some places follow it and others do not.

 

Call at 2PM or 3PM? Can anybody confirm?

 

2PM MDT, roughly 1 hour from now.

 

Blocked by the firewall. Wanted to join your chat.

 

Nothing really happened in the chat. I couldn't get to the call because work was too busy, so I joined the chat. No one said anything though, lol.

 

Its funny, really. I was all set to join the call and I got a call from another employer whom I have been wanting to work for. They wanted to do a phone interview, so I didn't get to join.

However, on a better note, I probably have this job if the big boss likes me. Got a face to face tomorrow.

Hope the call went well. Please give us the highlights if possible.

 

Congrats! Call went well from my perspective, but I'll let the fellas give the recap. After the first recap post, I'll have another post to follow up on some service issues.

 

Everyone was busy talking and listening, it was impossible to type and listen, at least for me.

Short version, while the cores of both the 580m and 6990m draw similar amounts of power (the TDP), the cards as a whole draw different amounts of power. The 580m draws significantly more as an entire card than the 6990m. The throttles are a response to this increased draw, both to keep the M17x within the specs for the MXM slot as well as to adhere to Dell specifications such as maximum skin temperature.

The M18x does not have the throttle because it uses different parts, the MXM slots specifically are different, and there is more air flow and cooling potential.

Other brands may not have the throttle because they are not following all the rules that Dell follows or because they have more room and cooling potential.

Basically, if you get 78C throttling at stock settings (so stock vBios), then there is going to be some contact information posted on these forums for you to use and you will be able to get a correct full set of replacement parts that you can install yourself or get a tech to install. However, there is no "wiggle room" and the engineer on the call was very clear that it is "maxed out" for Dell and the specs he has to adhere to. There will not be a throttle removal or a temperature increase on current M17x hardware because there is no room or ability for the engineers to do it.

Basically, where we are today, is that if you absolutely don't want throttling, then get a 6990m. If you get a 580m and have 78C throttling at stock you can get new parts to fix the problem. If you have a 580m and want to avoid throttles then you will need to use the solutions already covered on the forum. You can flash the vBios to remove the power throttle and you can use my HWINFO mod to avoid the 78C throttle, assuming your heat sink and everything is fully functioning.

Bill also put to rest the idea that re-pasting voids your warranty. As long as you don't break anything you are allowed to do whatever you want with your machine.

It was an interesting call, it is a bit disappointing that there is really no solution (beyond the ones we have already developed), but I understand that Dell is just doing the best they can with the specs they have. I still think the M17x is an amazing machine, and Dell really does care, as they have shown and continue to show.

 

I listened to the majority of the call, and after learning that FurMark should run without hitting the 78c mark I downloaded the app and gave it a run. is it normal for my card to be dropping from 20fps to 1 fps at the 63-64c mark?

 

For heat sink replacements, please send a PM to DellChrisM with your service tag, email address, ship to address (or where you want field service to happen), and daytime contact numbers. If you have more than one phone number you want to provide, please note which is the primary and which is the secondary. If you just want parts only, just give Chris your service tag, email, and ship to address.

I strongly urge you to literally watch the service provider like a hawk and politely interrupt him/her if you see something happening you don't like. You shelled out a lot of money for these systems, and it's our view you deserve to have your concerns addressed as best we or our service providers can at any point in time.

Thank you to all who joined the call. It was a pleasure. We'll do more in the future. John or Chris will probably run the next one, or we may bring the new external community manager (also a big gamer and super geek) into the fray. Special thanks go out to YodaGoneMad and widezu69 for raising awareness of the issue and putting the call together.

 

Thank you for the recap, Yoda. One of my main takeaways from the call was that if you are experiencing game stopping throttles then there is most likely something wrong with your GPU assembly whether it's the thermal pads, warped backplate or heatsink. Something is wrong. If you are seeing the occasional dropping of a few frames per second then your GPU may be performing to spec but your game is pushing it very hard.

If you are a benchmarking fiend wanting to OC and push everything to the hairy edge then the 580m may not be for you. I got it because I wanted a 3D screen and I don't regret it. The 3D is an awesome feature in my book.

At least that's what I got out of the call. Please correct me if I'm misrepresenting that.

 

That would be the power throttle we talked about at the start of the call. It kicks in at 63C and is avoided with the vBios mods, in terms of Dells perspective and engineering desires, yes, that is normal.

 

Thank you very much for the update. So, given that I do hit the 78 degrees Celcius throttle with stock vBIOS and clocks, are they giving us newer designed parts? For example, I know they could send most people a 6990 as a replacement, but with the my 3D screen, that's not an option.

Anyways, thanks all that were involved. It's awesome that this could be put together. As Yoda said, it's a bit disappointing, especially knowing that we don't really have the most powerful machines out there when that's what we thought we were getting, lol. I'm glad a company can reach out to us like this though.

 

Very dissapointing that there will be no fix to this. We all payed hard earned $$ for a computer that can't play games properly when it is designed to play games. Dell should be replacing the cards wuth an equivelant or better 600 series when they are lauched (that is of course if they create bios support.)

 

That is basically accurate, with one caveat. Dell is counting on the 63C throttle which many (almost all) of us enthusiast users have removed. With the power throttle removed even a perfectly functioning cooling system (like mine, which averages 65C during gaming on stock fans) can hit the 78C throttle.

There was that talk about the fact that what we see for GPU utilization is just a frame buffer, and that what really matters is the number of shaders used. Thus, you might be fine a lot of the time, then hit a scene that is suddenly using the majority of the 384 shaders and your temp spikes. It is these situations that my HWINFO fan mod really helps with, because it reacts extremely quickly to that temperature spike, where-as the engineer said the default fan profiles waits for the card to be above a certain temp for a certain amount of time before kicking up, and that wait might be long enough that you hit 78C and throttle especially if you have removed the power throttle.

 

They will be giving you a full set of the newest parts. Basically, you might not have the right parts or might not have them perfectly installed, and the 580m is cutting it so close to the specs that even a single loose screw on the heat sink will kill you.

I can confirm that with a perfectly installed cooling system and the mods on this forum the throttling is not really an issue. I have the vBios mod installed to eliminate the power throttle. I overclock 20%. Most of the time I just run stock fan settings and am just fine, the few times I hit 78C using my HWINFO mod fixes it. If you are not having a similar experience than you should get your cooling system replaced.

 

I do a lot of posting on the facebook page, and shared the info with them there, and wanted to post here as well, as I was trying to recap everything.

I hope I got everything needed, and didnt mess any of it up.

------------------------------------

Conference call took place with dell reps, and I have to thank them for taking the time to have the conference call. I wont go into too many details about it, because some of the topics discussed are a little beyond the scope of the Facebook page, and, frankly, I may not remember every detail that was discussed, but here are some bullet points.
(please excuse the typos, I'm trying to wrangle my peanut butter covered son and type this at the same time lol)



1. Repasting or doing any similar work on your computer WILL NOT VOID THE WARRANTY unless you break something. If you want to take apart every single screw and panel and part from your computer and reassemble it, repaste it, get out the paints and paint your name on it, or whatever...it will still be under warranty. The only thing to void it would be if you do something to cause damage to it, with the exception of that damage happening through instruction of a tech on the phone leading you through the steps.

2. The thermal paste on the 580m is better than mx4, arctic, etc. When you repaste it with something different you are actually increasing the issue.

3. The issue is all about cooling. Because of the smaller form factor, and the smaller fans, and the limited venting on the m17x r3 the heat increases more so than the m18x or desktops. This is a "duh" kinda thing, but explained more in the next point.

4. The heat is actually pushed to the edge on the system. The heat comes from the fact that the 580m draws more power than the 6990 (about 20% more). and therefore more amps and current control issues. This is a big concern for the i/o port of the card, the skin temps, and other temp issues. So the P0 state being the norm for high use, then clocks go down to P1 which is supposed to slow temp increase (essentially a 20% throttling), but not necessarily lower it, then once it hits 78C it goes into P8 state which then throttles it heavier.

5. Different directx versions, and driver versions can have a big impact. Seeing as shaders vary from one directx version to another, and different drivers can have their own set of variances which can impact use.

6. Ambient temps are tested around 28C (a little over 80F), so ambient temps should not be a big concern under than mark. However, ambient temps will still play a role, and can help reduce laptop temps.

7. The power throttling (movement to P1) takes place at 63C

8. (Quoted from YodaGoneMad on notebookreview.com) "There was that talk about the fact that what we see for GPU utilization is just a frame buffer, and that what really matters is the number of shaders used. Thus, you might be fine a lot of the time, then hit a scene that is suddenly using the majority of the 384 shaders and your temp spikes."

9. The m18x does have the throttle but doesnt have an issue hitting it like the m17x does. The 6990m also has a throttling point.


There are no plans right now, and probably will not be any in the future, to remove the throttling. It is there for a reason, mainly because of temp specification concerns and safety, but they should not be a continuous impact on your system, and therefore should be addressed if they are. Talks are in the works, however, about other issues such as with the bios, fan speeds, and other aspects that may directly or indirectly impact the throttling issue.



Ways to help alleviate the issue:

First off, understand that if you are hitting the heat throttle at 78C frequently, then you need to contact tech support. Need to make sure that your heat sink, fans, card, pasting, back plates, etc are all setup properly. A quick tech visit can assure this. There are assemblies available that are already pre-assembled and then shipped and can make it easier to just change out without need for repaste.

If you are hitting the throttle limit occasionally and not after every 5 minutes playing a game, some other things can help:

1. Check the basics. Increase airflow. Done have something blocking the fan outputs. Using some spray to clean out the fans, etc.

2. Custom fan speeds. Using HWiNFO64 you can set customer fan speeds that will be more aggressive than stock fan speeds. This can increase cooling performance.
Recommended set would be to set sensor to "Compal EC", reading the "GPU", with the following speeds:
0C=2000RPM
40C=2500RPM
50C=3000RPM
60C=3500RPM
70C=4000RPM
80C=4500RPM

3. Installing a new driver, vBios, or other software changes that may assist as being directly related or impactful on the gpu

4. Using a tool to disable powermizer (there are many different tools out there)

5. Using Nvidia Inspector or other tool to manually correct clock speeds and P states.

6. Making sure everything else is up to date. Other issues, like bad game installs, directx installs, increasing temps from other parts of the system, etc can be indirectly impactful
-----------------------------------



Edit: Forgot to note:

From the conference call today, with all that went on, its kinda hard to remember it all, but one other thing I wanted to point out. For those having hdmi-in issues, it is a driver issue. Whats happening is that when you plug something in to the hdmi-in port, the driver is shutting off the i/o port for the card since it believed there is another i.o option to override it. If you roll back to an older driver (before 285 I believe) you wont have this issue.

It is being worked on, but no resolution as of yet. Primarily in discussion with nvidia to correct.

 

Hi guys, I want to say a big thank you to all of you who have pulled this conference call together. Although I must admit that I wasn't expecting a major solution to the throttling problem, it was nice to find out the reasons behind the issue.

I have a question that I would like to be addressed if possible. From Bills post I understand that people who experience the throttling issue can request full replacement parts or get an field engineer to replace it for them. I would like to know if those parts would be new or refurbished? Thanks.