Nvidia GFX chips cooking themselves..

i'm sure we're all aware of the ongoing saga of some Nvidia 8xxx and 9xxx series chips unsoldering themselves due to chip balling that isn't able to deal with the thermal stress of being in a laptop.
even desktop parts are effected it seems.
HP have tried to remedy this with a bios update that runs the cooling fans faster and for longer. dell are having... erm.. talks... with Nvidia.
but with Apple recently taking NVidia as a parnter in chipsets and display adapters, maybe the problem is over?

anyway, i ramble on...

basicly, my faithfull Asus G1 bit it big time about an hour or so ago. with lots of graphical fireworks and glitches. i was gaming at the time, so got the best view of a 3D world going to blazes...

i can get it to boot safemode with a display off the VGA port, but there are blue lines all over the place, text is very corrupted, graphics are... well... interesting... then reboot normally, (still with glitches/lines all over the place), it goes to load the drivers, and fails. blue screen of death loading NVblahblah.sys, and reboots. the laptop screen won't display jack except grey bars all over the place. hence vga connection to external monitor.

thing is, i had a Geforce Go7700, so thought i was immune to the problems listed above with the Go8xxx/9xxx. but not so it seems. just a heads up.
trouble is, i'm about a week outside the 2 year warrenty ah well. sh*t happens.

 

well... its not a 9xxx problem A... and B... it CAN happened to any laptop... Not trying to harsh your mellow... just adding it in.

 

i would avoid nvidia for the time being

 

mainly the Geforce 8 series is prone to the faulty GPUs that overheat.

no issues have come up for the Geforce 9 series.

 

If your GPU has come unstuck, pardon the pun. A post months ago told of a guy who after being quoted $300 from HP to replace his mobo decided to try and re-solder his GPU and he got it working again, I'm guess if you have the equipment and the know how you wouldn't have much to lose trying this.

 

Any graphics card can die by over heating its just the g84 and g86 cores are more likely too.

 

All of a sudden, I'm in the mood for 'smores.

It's an embarrassment for g-chips to fail at the rates they have been. 1-2 years seems to be the average. The industry is well overdue for a radical redevelopment.

If you look on ebay, used nvidia laptop cards are going through the roof.

In the meantime, every time I game, I force my fans to high. It's my way of buying time.

 

If you're just a week outside of your warranty period, take a gander at seeing if you can still get yourself snuck in under the wire - call up customer service, explain that the first time you noticed something seriously wrong was two weeks ago, but that you were travelling at the time (or whatever, I don't know, be creative, but realistic) and so couldn't call until now. Be nice, sweet talk the person, and see if you can't get permission to send the thing in for repairs anyways.

 

Yeah if you bought something nice enough to have a 2 year waranty and it died that close I'd be calling up and trying to sweet talk, but being firm, them into fixing it. I'd imagine (or at least I'd tell them) I'd have to strongly reconsider buying their products in the future if they die just outside of waranty and won't stand by them. Any notebook can fail from thermal stress but the G8x and 9x parts used materials that were unusually prone to problems. If they absolutely won't warranty it you could try the heat gun reflow method that a lot of people have used as a last resort.

 

That sucks. I have to concure that you should talk to them about it. I remember reading some posts early in the the Enquirer's run of reporting on the situation that the 7 Series was also having issues. Seems nothing was ever offically said about the longjevity of those cards.

 

Nvidia just doesn't seem to get the picture...pun intended...especially since a graphics card cant display one! lol

 

Start at part 1 of The Inquirer's analysis. Then continue with parts 2 and 3.

I felt, at the time this was published, that it was written by someone who had a very good understanding of the problem. The past few months have reinforced my view. The safest operating condition is with the GPU kept cold.

While Dell and HP have extended the warranty on certain notebooks with nVidia GPUs, not all the manufacturers have followed this example.

John

 

According to the article, material expansion and contraction due to hot/cold cycling is a primary culprit, if we're to rely on the author's contentions.

Ultimately then, it is the improper choice and use of these raw materials and a lack of thorough R&D which is at the root of these failures.

Therefore the fault lies completely with Nvidia.

 

Yes, but the point is that they made several material choices wrong, combined with a layout that couldn't be adapted.


Really, they basically decided to play it safe with the underfill as well as the high-lead balls. I doubt anyone would have called them on their actions before the results came back. If they had used the new eutectic balls and run them at their current limit they would still be failing left and right after a few years anyways. The cards, when they were made, were required to turn out more performance than they could be expected to with the materials available. To me the only real problem was a lack of testing, but they probably did it all in desktops with huge heatsinks: thus ignoring the area where the problems happen: laptops being moved around when they are overheated.

Its like when you set a table for a meal. One time you drop a plate or a glass and it breaks. Could it be prevented? Yes. Is it your fault? Yes. But eventually something is going to go wrong. The real problem here is that nVidia isn't fixing the issue. They should be offering extended warantees (to cover normal usage, at least 3-4 years) and replacements to people who can easily swap cards (MXM).

 

Examining and auditing the problem from a technical standpoint is one thing. But there is a larger issue going on here as well.

The themes here are product liability, pushing the limits of technology and a lack of competition.

Manufacturers have long been held to a standard of strict liability when it comes to public safety. People have lost arms and legs in combines, the Ford Pinto exploded on rear impact collisions and babies have suffocated because of improperly designed cribs; these are some example of strict liability. Remember the laptop battery recalls of late? If Nvidia produced products of this kind, there would be massive and immediate recalls along with gov't intervention.

Of course, g-card mfg's aren't held to that standard but they do face civil liability if they produce products with massive premature failure rates. Two years is premature by any standard. Nvidia must address this problem and at the very least begin manufacturing low cost replacements to support it's customers. The push of technological expectation is not an adequate excuse for producing defective products.

The lack of viable competition in the industry also contributes to this kind of outcome. As a single supplier of high end graphics for Dell laptops, what choice did customers have? Quality, reliability, word of mouth and public perception would shape the industry. Not so in this case.

At the very least, Nvidia/Dell, Nvidia/(your laptop mfg), must begin the aftermarket manufacturing of low cost replacements, with perhaps a prorated discount for their failed and defective GPU's.

Seriously, can Nvidia / Dell honestly say that so many customers have abused their GPU's?

.

 

I hope someone from Nvidia and Dell take note of the recommendations above and pass it along to their higher ups.

I have one of these cards in question and luckily it's still holding up. Knock on wood. Then again, at these failure rates, as the availability of used replacements decrease prices will further escalate.

A bad card renders the entire laptop useless.

Dell / Nvidia are obligated to begin production of low cost replacement GPU's starting with Quadro 2500m/GeForce go 7950gtx series and offer them at marginal profit margins. These would incorporate the 1500m/7900gs defects as well.

A short 20,000 unit run is very much within their capabilities and obligation.

.

 

I read all three parts of that article, not to mention this one as well (which describes how the 55nm chips are affected)......

Seems like a good chunk of the laptop-using population is going to be in for an ugly surprise. Adding insult to injury is that nVidia refuses to identify what specific chips are defective (mentioned in the article).....meaning whether you order from a manufacturer online or walk into someplace like Best Buy, Fry's etc., you're playing a game of Russian Roulette--not knowing whether the GPU in your chosen machine will bite the dust (and if so, whether it'll die sooner rather than later)

 

This has been a problem with nVidia for a long time. I had the same thing happen to my GeForce 440 Go some 5 or 6 years ago, and it was a common problem at that time as well. It may also be a problem for ATI, but I've just never experience it with them.