nvidia heat issues
nvidia chip defect
Nvidia has uncovered a problem with some older graphics chips that shipped in "significant quantities" of laptop PCs, the company said Wednesday.
Nvidia hasn't determined the exact cause of the problem but said it relates to a packaging material used with some of its chips, as well as the thermal design of some laptops. Modern processors generate considerable amounts of heat.
To tackle the problem, the company is releasing a software driver that will cause system fans to start operating sooner and reduce the "thermal stress" on the chips. The driver has been provided to laptop makers directly, said Derek Perez, an Nvidia spokesman.
Nvidia will take a charge against second-quarter earnings of US$150 million to $200 million to cover the expected cost of repairing and replacing the products, which include graphics processing units and media and communications processors. It didn't say specifically which of its products were affected.
The products have been failing in the field at "higher than normal rates," Nvidia said. In a filing with the U.S. Securities and Exchange Commission, it said it was talking to its supply chain about getting reimbursed for some of the costs.
The company also had other bad news on Wednesday. It said it was lowering its revenue forecast for the second quarter due to pricing pressure and delayed product ramps. The company now expects revenue to be between $875 million and $950 million.
http://news.bigdownload.com/2008/07/03/ ... cial-news/
http://news.bigdownload.com/2008/07/03/ ... edictions/
http://www.reghardware.co.uk/2008/07/03 ... st_glitch/
http://www.theregister.co.uk/2008/07/03 ... st_glitch/
Nvidia issued some somber news for shareholders today, revealing a financial forecast cut short due to slowing sales, a delayed ramp for new product, and a hefty payout due to faulty laptop chips.
The graphics giant said it expects to pay between $150m and $200m to cover warranty, repair, return, replacement and other costs for defects in certain laptop GPUs (graphic processing units) and MCPs (media and communication processors).
As of yet unspecified notebooks using previous generation GPUs and MCPs manufactured with a particular die/packaging material set are failing at higher than normal rates, the company said.
A filing with the US Securities and Exchange Commission states the number of notebooks shipped and sold with the defect were in "significant quantities."
"While we have not been able to determine a root cause for these failures, testing suggests a weak material die/package combination, system thermal management designs, and customer use patters are contributing factors," the SEC filing states.
An Nvidia spokeswoman said the company won't reveal which laptops are affected by the problem out of consideration for its customers (that would be the vendor customers, not the end-user customers obviously.) She described that the materials used in this particular die set are not robust enough to handle the thermal stress in certain laptop configurations.
Nvidia intends to fully support vendors in their repair and replacement of impacted boxes. So, we guess that's fire suits and apologies all around.
"We continue to not see any abnormal failure rates in any system using Nvidia products other than certain notebook configurations," the filing states.
"However, we are continuing to test and otherwise investigate other products. There can be no assurance that we will not discover defects in other MCP or GPU products."
Nvidia has developed a band-aide software driver that will cause the computer's fan to start when the system powers on, giving it an instant cool down. So, if your computer starts buzzing after a system update, you'll know you're an unlucky recipient. Share your grief with us. We'd like to know what systems are among the afflicted.
Adding to the good news, the chipmaker noted that second quarter revenue is expected to be lower than previous forecasts because of poor sales worldwide, the delayed ramp of a next generation MCP, and lower GPU prices.
"This has been a challenging experience for us," said Nvidia CEO Jen-Hsun Huang in a statement. "However, the lessons we've learned will help us build far more robust products in the future, and become a more valuable system design partner to our customers."
It's also hoping to avoid public confrontation with the likely source of the chip issues — its foundry partner Taiwan Semiconductor Manufacturing that makes most of the company's kit.
When asked if TSMC was responsible for the material defect an Nvidia spokeswoman replied, "We aren't saying 'yes' to that question now."
"We're being careful to take our share of the responsibility rather than throw anyone else under the bus," she said.
Nvidia now expects sales for the quarter ending July 27 to range between $875m and $950m, which would be below analysts' estimates of close to $1bn in revenue. Investors responded to the news in after-hours trading — dropping Nvidia shares nearly 22 per cent at time of publication. ®
According to The Inquirer, the heat issues affecting “significant quantities” of older mobile GPUs affects all NVIDIA G84 and G86 GPUs which are used in GeForce 8400M and 8600M graphics cards.
Now you can take the Inquirer with a large pinch of salt if you wish, but last week when news broke of this issue I followed the trail of destruction that that led me to conclude that the GeForce 8400M was a likely suspect. More digging, along with news of an update from HP made me add the GeForce 8600M to the list of suspects. If some are affected, it’s not a big leap to assume that all (or a large proportion) are affected, making this a very big problem indeed for NVIDIA. Want to dig the hole even deeper? Well, consider that these GPUs have also been used in countless notebooks, including Apple’s MacBook and MacBook Pro lines. If this really turns out to be this big, then $200 million isn’t going to go very far.
The way I see it this kind of problem was almost inevitable. Both NVIDIA and AMD/ATI are responsible for pushing higher and higher voltages through the silicon and relying increasingly on the coolers to move more and more heat away from the GPU. Any weakness in the thermal system (joints, thermal grease, epoxy, fan performance …) is going to lead to problems, and even a small design flaw is capable of becoming a huge problem down the line.
Until NVIDIA come clean on this issue (and according to one expert I’ve spoken to, it could take months to fully get to the bottom of this problem) uncertainty will cloud the company.