"fabulous node" could mean a lot of things. It could mean 28nm is great in theory, will be great, great for what it promises. It doesn't automatically mean Nvidia is having great success with it currently.
"fabulous node" could mean a lot of things. It could mean 28nm is great in theory, will be great, great for what it promises. It doesn't automatically mean Nvidia is having great success with it currently.Giving Praise to TSMC.
Actually, they're everything.
Correct me if I am wrong, but AMD is not die-harvesting their 7970 GPU's, correct?
Correct me if I am wrong, but AMD is not die-harvesting their 7970 GPU's, correct? They are fully functioning chips with all designed-in features active and yielding.
That tells us the functional yields of 28nm are not limiting AMD's market plans for 7970. Price may still be based on supply/demand that is supply limited simply because TSMC is 28nm capacity limited.
I don't know, there are a lot of possibilities here. Then again, had the 40nm sitation not played out as publicly as it did then I would not have considered it possible for Nvidia to miss the target on DFM as badly as they did when AMD managed to nail it as well as they did.
"fabulous node" could mean a lot of things. It could mean 28nm is great in theory, will be great, great for what it promises. It doesn't automatically mean Nvidia is having great success with it currently.
Glen Yeung - Citigroup Inc, Research Division
Two questions. The first one is maybe just some clarification from you, Jen-Hsun. I think I'm hearing that GPU demand is being impacted by a combination of lower PC demand from hard drive shortages, hard drive ASPs squeezing out GPUs for the bill of materials and then a shortage of capacity from 28-nanometer. Can you just give us a sense -- maybe rank those in terms of what's most impactful to sales and diverse lease [ph]?
Jen-Hsun Huang
28-nanometer shortage. I could use more 28-nanometer wafers today.
50% of your posts don't even make sense. And another 25% amount to reiterating word smithing used by some CEO.
Which one the above is, I can't even tell...
What point are you trying to make? Nvidia said they need more wafers. So? I don't see one Kepler based GPU I can put in my shopping cart, did someone buy them all already? The need for more wafers strongly suggests they are using the wafers for something, which begs the question. When is Kepler hitting the market? Why didn't Nvidia give any granularity here?He was giving praise to TSMC to help solve the lower than expecting yields nVidia is having.
They're having less than expected yields but that's not the real problem but having more wafers is. There was a wonderful question in the call:
This horrible situation as it is known by nV at this point is expected to impact their margins all the way down to 49%, that's only ~10%(4% absolute) higher then AMD, and increase their operating expenses by an awe inspiring 4%......
As a business report- yield issues explain why operating expenses are going to increase and why margins are going to see a decline. nVidia could still produce millions of GPUs with yields below expectations, it would simply impact their margins.
That tells us the functional yields of 28nm are not limiting AMD's market plans for 7970. Price may still be based on supply/demand that is supply limited simply because TSMC is 28nm capacity limited.
The issue that you're missing is that they have a given amount of wafers to work with and, by the looks of it, quite large obligations to OEM that they absolutely will service before the retail market if they do not wish to face unpleasent consequences in the future. When chips are short, desktop discrete will be the first to get the shaft, and that is the part of the market that most of us here are intersted in.
Also, Sontin, how would you explain the reasoning behind shiping the Gtx480 with a part of the shader cores locked?
What point are you trying to make? Nvidia said they need more wafers. So? I don't see one Kepler based GPU I can put in my shopping cart, did someone buy them all already? The need for more wafers strongly suggests they are using the wafers for something, which begs the question. When is Kepler hitting the market? Why didn't Nvidia give any granularity here?
BTW, lower yields mean the need for more wafers because of waste, so for you to say, "less than expected yields not the real problem" is silly.
That would be your problem -- if you actually read more instead of ignoring so much -- would help
Or perhaps AMD, too, is suffering with poor yields but decided not to talk about it in public. They don't need any more bad news. The 'uncompetitive' pricing of latest generation AMD GPUs tells its own story. People get so carried away with the usual AMD versus nvidia BS that they don't stop to think for one moment that the real issue is with TSMC.
Makes you appreciate what Intel has achieved in the past few years. A flawless shrink to a full node every two years. All this, while leading the charge to smaller processes well ahead of others. Incredible. Either TSMC are incompetent and suck more than we ever realized, or Intel's foundry engineers and scientists are aliens.
If you desire to lock onto the yields, hey, one can and there is guidance on how they will effect margins to around 49.2 percent. But, for me, the problem is more-so in the sheer amount of wafers so nVidia can try to meet demand when they launch.
It's all related. If they had 80% yield (hypotherical), they would not need as many wafers. If they are throwing away 80% of their production (even after binning), they will have a 'wafer shortage'. Two ways to say the same thing, it's all about yield. This is where Intel has been killing everyone.
This was reported during an earnings call. If they lie, they go to jail. It is criminal.
If you spent less time rushing into threads to play the white knight and start in on your ad nauseam financial breakdowns would make it easier to see some sense in half your posts.
hmm i never heard JHH go to jail when they unveil woodscrew fermi, its was blatantly lie.
I know using exact quotes and data from the companies surly doesn't make much sense compared to rumor mongering, unnamed sources, conjecture, speculation and Internet Lore.
The demo they showcased at GTC was using Fermi hardware.
Funny how you tend to see it. The majority of your posts leave the impression that you are to nvidia as Fox News was to the most recently departed US administration; an echo chamber without any sort of personal assessment, just straight assimilation and echo.
You charged right into this thread and started playing damage control when no one was taking a strip off nvidia, just talking about this article. You'd do well to assess why you feel so compelled to downplay or put positive anything in these forums that comes across as even mildly negative towards nvidia.
It's quite amusing.
so just tell me is this image below is real fermi hardware ??
I'm listening to the conference call now (thanks for the link, not) and JHH said several times that they are working with TSMC and that yields will improve. Which is stating the obvious to say the least of course they will. AMD is somehow able to ship 28nm in good quantities, so did AMD get all the good wafers somehow? JHH said that hardly anyone is shipping 28nm products, even though their direct competitor has been shipping them since last year. Quite disingenuous on his part.If you desire to lock onto the yields, hey, one can and there is guidance on how they will effect margins to around 49.2 percent. But, for me, the problem is more-so in the sheer amount of wafers so nVidia can try to meet demand when they launch.