Never thought the day would come where I'd hear someone say this.
Never thought the day would come where I'd hear someone say this.
Don't you know you can always find someone to say what a person or group wants. There will always be an individual or a group to maintain an idiotic position.
Never thought the day would come where I'd hear someone say this.
Tegra 3 is a quadcore ARM CPU with a die size of 80mm squared. For comparison purposes, the ipad 2's cpu, A5, is a dual core CPU with a die size of 120mm squared. Nvidia manufactured Tegra3 on 40nm instead of 28 because 1) it was a known quantity and 2) it allowed them to be the first company in the world with a quad core ARM cpu.
Well the bottom line is Nvidia is talking about worse than expected yields, AMD is shipping product. The yields are obviously not bad enough to prevent AMD from shipping 2 (soon to be 3) different cards in the space of a couple of months. The optics just come across as not so good for Nvidia, it makes it sound like Nvidia is blaming TSMC for lack of Kepler products, whether that is the case or not the inference is quite easy to draw.
AMD CEO never said yields were good, he said the yields were exactly as expected. JHH said they were less than expected. I wonder if both are getting the same yield rate, yet AMD correctly calculated it and based their products on that while NV (once again) did poor headwork and now their plans are foiled.
nVidia is not blaming TSMC for a lack of Kepler. They never expect to ship Kepler in their financial Q4 since mid 2011. They are now in the ramping process and they see that their yields are lower than expected. That's normal when you start the production of the first products on a new process.
Well the bottom line is Nvidia is talking about worse than expected yields, AMD is shipping product. The yields are obviously not bad enough to prevent AMD from shipping 2 (soon to be 3) different cards in the space of a couple of months. The optics just come across as not so good for Nvidia, it makes it sound like Nvidia is blaming TSMC for lack of Kepler products, whether that is the case or not the inference is quite easy to draw.
I remember reading that at 40nm AMD looked at the projected leakage figures that TSMC were providing and came to the conclusion that they were overly optimistic so they designed around it.
Missing expectations is now considered "normal"?
Anyway, I'm sure there were numerous problems with GF480 but JHH is always quick to blame everyone but Nvidia for all problems...
Never thought the day would come where I'd hear someone say this.
One can generally expect yields to be lower early in a new process or node. However, at any point in time, if yields are lower than expected, it says something about your ability to make realistic projections. Fermi also appeared to suffer from overly optimistic expectations. Is there a trend here?nVidia is not blaming TSMC for a lack of Kepler. They never expect to ship Kepler in their financial Q4 since mid 2011. They are now in the ramping process and they see that their yields are lower than expected. That's normal when you start the production of the first products on a new process.
Never thought the day would come where I'd hear someone say this.
there wasn't supposed to have a demo today?
One can generally expect yields to be lower early in a new process or node. However, at any point in time, if yields are lower than expected, it says something about your ability to make realistic projections. Fermi also appeared to suffer from overly optimistic expectations. Is there a trend here?
Again, just excellent info; this is very insightful. It would seem to me then that NVIDIA will have difficulties in both yield arenas considering their approach. If functional yields are low for everyone because TSMC's 28nm process isn't doing so well, that's bad. Couple that simply with a physically bigger chip, I imagine yield plummets. If they are facing a parametric issue like they did with Fermi, then good luck seeing anything until there's been several respins.
If we're to believe the rumors/hype that Kepler was focused on power efficiency, maybe that will help so they can get some kind of product out the doors at a reasonable TDP (although they'll miss their performance target). I just find it funny that two years later, it doesn't seem like they learned from their mistakes with Fermi. I know designing chips takes an awful long time, but one would assume they wouldn't have placed as much faith in the process shrink this time around, although it seems they did. Where this looks even worse for consumers is that TSMC reportedly (or is it a rumor? I forget) isn't giving them the same pricing as Fermi, and they'll pay by the wafer like most. I imagine TSMC took a major hit when Fermi was released, and they won't repeat that mistake again.
"The number of tape-outs on 28nm compared to 40nm has been triple at the same stage of production."
According to the official of TSMC, the company already has 90% of the 28nm foundry market.