Alright its fact now - 8800 Series GPU'S are defective - what does that mean for us with 9800GT?

You tell us....has your graphics card overheated? What are your temps? People should remember that the gpu's don't overheat till they get into the 97-110c range. I don't think I've read anything about a 579X going over 72c.

 

Hmm, the thing is, I have this desktop 8800GTS (640MB) for..since release, 1.5 years? and it is in a desktop PC without the absolutely best airflow and in a very hot room for at least 4 months (room temperature over 34C) and I haven't encountered anything that would lead me to think there is anything wrong with the card. The article says that desktop versions are affected too. I just don't know. The 8800 series for desktops are like 2 years since release (I think), if there was such a serious problem that will not take long to appear as they claim, why haven't I seen anything reported or hinted at all the forums I check etc for the past 2 years? The GPU temperature has been as high as 80s (C) in the past, though thanks to fan programs to set the fans at 90% speed the temperature doesn't go higher than 74C at extreme load.
The mobile version of the 8800 is out for several months too, correct? I am new here, but how many people have posted here (or anywhere that you know for sure) that had such problems?

Last question, IF it is true, any idea if the 9700m series is affected too?

 

I don't think the issue is that it runs any hotter. The problem is that the materials in the interface between the die and the packing, ie., solder, does not stand up to repeated heating/cooling cycles. If my memory serves me correctly, an article a year or so back talked about the problems of lead-free solders under RoHS and one of the issues was heat cycling failure. I will try to find the article...

 

So BlueMak, having such a high ambient temperature may actually be protecting your card from having such radical thermal cycles by keeping it at a fairly high temperature.

 

That's good to know, although I am not so much worried about this but for my future notebook.

BTW, isn't this issue similar with what the Xbox 360s are/were having?

 

why are you guys worried so much? Just use the GPU a lot until it break within the first year and get a replacement...

 

I don't know for others, but I will also use the notebook for every day important to me work and I won't be able to just do nothing till it goes for repairs/replacement. Even one day can be one day too many.

Plus, to buy something that you know it will break down, is kinda...not smart.

 

with all your guys zalman coolers and the jet engine they installed for a fan i'm not sure the 579x owners have much to worry about....

 

that's why i'm refocussing this thread to the M860TU not the 5796.

 

LOL I was referring to spexc31

 

So just to clear things up...

Those of us with 17" laptops such as 579x/9262 should theorhetically unaffected by this due to superior cooling solutions?

 

The problem is (mainly) caused, as people have said several times, by heat cycling weakening the solder used on the chips. So it's not so much a question of how hot but rather how quickly it gets hot and then cools down again. As long as you can prevent large, rapid fluctuations in gpu temp then your card should be fine.

 

Good thing mine will be largely on most of the time.

 

So with BETTER cooling in our large laptops we heat cycle faster which is bad?
OR
Because the big laptops have better cooling the solder doesnt get as hot and thus the heat cycling is to a smaller degree and thus less damaging?

Any electrical engineers here ti chime in?

 

Well, for what it's worth, the 8800M GTX in my 5793 regularly hits 80C and higher while playing games; this is on a hard wood surface with cleaned vents, and it's been doing it since I first got the laptop. I thought the problem had been solved by switching drivers, but it came back again. XoticPC support basically told me I had nothing to worry about, but this topic has piqued my curiosity once again.

 

gosh this is giving me a headache and I haven't even received my laptop yet ...

 

Your machine will blow up and burn the earth. Now get an Alienware m15x!

 

You know what? I'm just not yet ready to buy that the 8800 is defective. Before the actual users, in the real world, are reporting problems (outside of the m15x chassis), I'm to believe that there's something wrong? Sorry, I can't do it. There's going to have to be a pandemic level of reported hardware failures before I even bat an eye at these "reports". Even the 8400/8600 haven't seen high failure rates, but so many people are in full "the sky is falling" mode.

I can't and won't operate in that manner. Sincerely, I'd like to be corrected, if I'm looking at this the wrong way.

 

http://www.tomshardware.com/news/nvidia-g84-g86-chips-overheating,6121.html

It seems that maybe only the 8400s and the 8600s maybe be affected, hopefully.

 

heres a post from engadget, it says 'most of 9800 suffixes too'

http://www.engadget.com/2008/08/12/nvidia-g92-and-g94-gpus-failing-too/

 

i wonder if tomshardware tested out the g92 core(8800/9800s). not mentioning it doesn't mean their not defective necessary yea?=\

 

notice those are desktop cards?

 

but it says 'several mobile flavors of'... i might just be reading it wrong :S... anywhoo i dont think sager would have this problem much as it cools way better than other laptops

 

I think the problem should be fixed by now, check out this article and I like this sentence.

"NVIDIA says it's talking to its suppliers about being reimbursed for the faulty parts." --- My question is are we consumers also getting reimbursed?

http://www.engadget.com/2008/07/02/nvidia-says-significant-quantities-of-laptop-gpus-are-defectiv/

 

I think any GPU that used a bad batch of solder is at risk. That does not mean to say that the risk will turn into a problem, it all depends on whether the cooling system is set up to minimize the thermal cycling.

Here's the article I read. Most of it is on something unrelated (still interesting), but the important part is near the bottom.

 

my 8800 GTX card broke in some way. I don't know if that is related to this. I got it replaced with the warranty though. It was being used normally, no OC or anything like that.

 

I'm sorry, but this reeks of pure FUD. "Unspecified board partners... going bad at high rates." Wouldn't we, the enthusiast community, know about these high rates long before the Inquirer or Tom's Harwdware? These cards aren't like IGPs, where everyone and their mother has one. These high-end graphics cards are owned by a very small and very outspoken percentage of the computer owning community, and the news of bad product spreads like fire ants on watermelon rind. If we're not complaining about a widespread issue, it doesn't exist. Period.

 

Not that I, in any way, believe The Inquirer's almost sourceless claim - but when a company's under suspicion of something, it isn't wise to quote their press releases as fact.

Not that it's important in this case - you just set of my skeptic-o-meter.

 

If there really was a problem with any of the 8xxx or 9xxx series video cards in the Sager, PowerPro or Force models it would be all over these boards.

Please understand that it is nVIDIA that makes the video chip, but it is the ODM that makes the complete video card and thermal solution, so when it comes to Clevo, Compal or MSI laptop shells the quality of the thermal solutions is made with heavy use in mind, without the cost cutting mentality that you find in the major national brands.

Bottom line, if you don't see gobs of reports of failed cards here it is because it isn't happening. I don't mean the occasional failed video card, that is going to happen to any card, what I mean is the large percentage of failed cards that the major national brands are reporting.

We provide a standard 3 Year Warranty w/Next Business Day Shipping with all of our PowerPro laptops, and Sager provides a 1 year warranty with 2 or 3 year warranties offered if you want them. If we were the least bit concerned about this, we wouldn't be selling models with these cards in them. In addition, our warranties are backed by major international companies that are also backed by MSI, Compal and Clevo.

 

i don't get the difference between the 8800m GTX and the 9800m GT.. is it just the temps? are the performance actually the same? =/

 

Same exact card just renamed.

 

from the newest info, they are not the same.

 

Alright! I am glad they are not the same, now I feel better ordering my new laptop :laugh: (wish the color is not pink)

 

Do you have a source for the "newest info"?

 

don't get it.. same, why rename? confused =/

 

I am not really sure myself but I think it's becaause they released thier new 260 and 280 cards, but I could be wrong.
From everything I read the new 9800M GT cards are the same exact thing as the "old" 8800 GTX cards.
It still is all very confusing to me.

 

@IKAS V

yes, it is the same card with different driver.

Please look: http://www.notebookcheck.net/Mobile-Graphics-Cards-Benchmark-List.844.0.html

 

I'm with Donald. What's the difference?

 

Maybe only the smaller manufacturing process?
Something else?
It shouldn't have the 112 SP of the 8800m gtx (even with 16 being lasercut), and have only 96.

 

geez, the perfect monent ati to release a whole new batch of Gpus and overwhelm nvidia

 

I would like to have the 9800M GT in a D901c, since 2 mounth ago. But european reseller starts on 20.aug.2008 with this card, earliest. I wrote very many mails and the result is: Clevo is unsafe (?) to give warranty for 9800M GT 512MB and 9800M GTX 512MB and 1GB in D901c chassis.
I wrote eurocom, malibal, pctorque, clevo.com.tw, notebooktotaal.nl, nexoc.de, notebooksbilliger.de, hoh.de and many others.
Nexoc canceled all D901c chassis without replacement, concerning heatproblems in the long run with 9800M GT .
I wrote eurocom and confronted they with this statments.....no comments at this time. Only a delay from 3-4 weeks was confirmed.
Any high qualitiy resellers wrote me follows: Any Resellers hope, the 9800M GT from other Clevo`s fits to D901c and this is a mistake. Nexoc and pegocom.de wrote me that.
I can send this email to admins in this forum, if you would to have this for check.
Comments?

(Sorry for errors in orthography, I`am learning!)

 

Then you might as well not buy anything at all because eventually everything breaks down.

Also, "defective" is a relative term, given that everything breaks down eventually. Since the longest warranty available, to my knowledge, is three years, I would hazard a guess that the typical life-span of the primary components in a laptop is typically estimated at about three years plus a little - i.e., probably something like the concept of half-life for radioactive nuclei; three years plus is the median time to failure for components such as GPUs.

On that basis, if your GPU doesn't fail within 3 years, ipso facto it isn't defective, even if, with a better soldering solution, it would have lasted longer than it actually did. It may be that the various vendors involved - NVidia, Clevo, Sager, etc, may decide to offer a trade-in program of some sort if they can more specifically identify the batches of botched GPUs; however, that's not something that I would count on, and may not even be something they can be forced to do. If you're really concerned, the best thing to do is to buy a three-year warranty.

Also, the problem, as one or two posts have pointed out, is rapid, high delta changes in the GPU's overall temperature; i.e., the temperature of the PCB plus the chip - focused, of course, on the solder joints where the chip meets the PCB in this case. So, what needs to be avoided is repetitive, large changes in GPU overall temp in a very short period of time. The mere fact that it gets very hot is not a substantial factor (although it is a factor to some degree, pardon the pun ).

The best way to engage in preventive maintenance is (i) first and foremost, follow Gophn's Cleaning Guide religiously, (ii) never operate with anything blocking any of the vents - that means, for example, don't use it on your lap (come on people, this is a DTR after all, emphasis on Desktop), and (iii) don't aggravate the situation by intentionally causing high, rapid changes in temperature by going directly to a graphics intensive use right after a cold start, or by doing a complete power shutdown right after extensive graphics intensive use.

On this last point, think of it like doing exercise - first, you have to warm your muscles up in order to prevent injury, and second you have to give your body a chance to cool down properly before stopping all activity. Just so, you have to warm your GPU up properly on startup, and you have to let it cool down before turning it off.

For example, right after you do a cold start, do a little surfing, open multiple windows/browsers, grab one and wave it around on screen - that'll give the GPU a warmup by having to redraw the screen, but nothing really intensive. Then, when you've finished gaming, for example, close the game out, do a little surfing, and then let the system sit and idle for a few minutes. Once it's done that, then you can power it down.

 

Ok, time for my fast car reference....
When you start up the new BMW M3 the tachometer has a variable redline that starts a a relatively low rpm until the engine warms up. Then once everything is up to optimum temp the redline moves to its normal position. The M3 doesnt let you push it over the edge on a cold engine, but once it warms up you can really fly!
So I guess those little temp monitor apps can be our engine temp guide to alert us to when we can start turning our clevos up to 11.

But whats a good optimum operating temp?

 

Probably anything that's within spec and below the ceiling threshhold for throttling and/or potential damage from heat alone; otherwise, benchmark off your own GPU's performance. The thing to keep in mind here is that, other than the spec ceiling temp, the actual operating temperature is less important than the delta on temperature changes.

 

Monitor temps,clean regularly

 

So should I order a M860tu or what? When will they ship?

I have a thinkpad r51, but it's time is done.

I won't be able to afford (educationally & cost-wise) shipping a laptop back and forth for servicing starting next month.

 

I disagree with the warm up part, but I am sure you do need a good cool down.

I believe we are talking about a bad soldering here, which is a material property. I said I disagree with the warm up, because doesn't matter how rapidly you increase the temperature the material melts at specific temperature, so the time it takes to reach that melting temperature shouldn't matter, unless it take a long time for the fans to kick in and ventilate adequately. I believe cool down matters is because of the fans again, if you shut down your laptop your fans will stop too, and just after doing graphic intensive stuff your GPU is going to be pretty hot and stopping fans would stop ventilation which would also cause a heat build up inside the GPU and could result in failure.

 

I think it sound a bit strange with letting it cool off with the fans, wouldn't it cool off faster with fans and not slower like it would in room temperature...

And if the soldering is bad, wouldn't that just mean that a broken computer with this problem just has a bad connection and that the GPU itself should be fine? which should be possible to fix by yourslef if you know how to solder... or is it too small to get to?

Or am I wrong?

 

wait... doesn't anyone else find what paladin said interesting?

Im almost reading that as its not nvidias fault largely, but the manufacture yea?
the reason why cards are melting down is cause the manufacture like dell are doing a poor job implementing thermal solution and etc.
so theres no problem with the cards nvidia gives itself?

 

Well on the 8600gt nVidia has said that it is their own fault but also the manufacturers that has too bad cooling...