Fermi mass production pushed to May or later

[Matthew DeCarlo]

After unveiling a fake Fermi card, shrugging off AMD's lead, and months of uncertainty, Nvidia has provided a few details about the mass availability of its next-generation graphics cards. During a conference call, Nvidia CEO Jen-Hsun Huang revealed that Fermi would hit "full stride" in the second quarter of fiscal 2011 -- or May to July of 2010.

That timeframe will bring several Fermi-based GeForce, Quadro and Tesla products spanning many price ranges, but the first cards should arrive before then. The company is expected to start shipping the highly-anticipated GeForce GTX 400-series graphics cards in the first quarter of fiscal 2011 (January to April, 2010). Other Fermi cards could also debut in current period, but nothing is official at the moment.

While details are limited, Huang did say that the speed of the transition to Fermi-based products depends on the supply of 40nm chips. We thought TSMC's 40nm production issues were resolved, but it seems Nvidia is still being underfed. Huang estimates that the shortage has cost Nvidia between $100 million and $200 million.

Permalink to story.

https://www.techspot.com/news/37963-fermi-mass-production-pushed-to-may-or-later.html

 

[Guest]

It seems you may have your years messed up. Might want to check when you use 2011 and 2010.

 

[Matthew DeCarlo]

Nope. The years are correct. Thanks anyhow.

 

[Anshrew]

What happened to the GTX300 series cards?

I don't like this new naming scheme, it's messed up.

whats wrong with a Geforce10000 and Geforce11000 series.

 

[Guest]

It seems you don't understand business finance. You might want to take a finance or accounting class.

 

[Guest]

Check out this link to SemiAccurate:

http://www.semiaccurate.com/2010/02/17/nvidias-fermigtx480-broken-and-unfixable/

 

[thatguyandrew92]

Woah, this really sucks.

 

[dividebyzero]

Check out this link to SemiAccurate:

http://www.semiaccurate.com/2010/02/17/nvidias-fermigtx480-broken-and-unfixable/

Hardly surprising that Charlie has his panties in a wad - He's been crying ever since nVidia done him wrong.
You might want to bear in mind that this is the guy that predicted:
ATI R600 graphics cards to have inbuilt sound card
That the GTX275 doesn't exist
The nVidia G80 (8800GTX) shaping up to be “patchy” (FYI the G80 is considered to be one of the best GPU's of all time)- not only that but the nVidia 8800GTX borked and has no SLI capability
...and that ATI's R600 and the 8800GTX would hit retail at the same time
ALL – not some- but 100% of nVidia G84 and G86 GPU’s are bad
That the GTX260 to cost 50% more than HD 4870, HD4870 1Gb beats GTX280 (which is slow and hot….although not as slow or as hot as the HD 4870 presumably)
and of course how could we forget his rant about Windows 7 RC being slow-w-w

Big pinch of salt with this guy's writing.

 

[Per Hansson]

dividebyzero, R600 does have dedicated silicon for the sound support, so he is right on that point

The GTX 275 issue was that a special run of 280 cards where made from the GTX 280 ASIC and then sent to review sites, what then went to the retail and OEM channels was the GTX 260 ASIC cards

He is not saying that it does not support SLI, he is saying the release drivers where buggy and crashed all the time, and that this would later be possible to fix (which it to some extent was, but do a Google search for nVidia SLI problems and you will probably be able to spend the rest of your day reading on problems people have with it )
As for the HDCP issue that is real, 8800 series does not support Dual Link HDCP output, and if you have a 30" monitor I think that HDCP support would be on the table so you can run the thing at full resolution and not have to downscale to 1080p

The failure rate of the nVidia graphics cards due to bumpgate is a known fact, or you think nVidia filed a K8 just because it's fun to give their customers $200 Million in extended warranty replacement parts? (I've got a Dell M1710 laptop here with exactly that issue)
And it is heat related, so the better cooled the card is (the less difference there is when it is cold and hot) the longer it will survive (which explains why there are less problems on desktop systems and many more on cramped laptop systems with poor cooling)
http://www.nvidia.com/object/io_1215037160521.html

I'm not saying Charlie is someone I like for his bashing and generally poor journalistic writing, but especially the nVidia bumpgate fiasco he did a great work on which the normal press did not dare to cover

 

[Guest]

Yea the years are correct.
Anyway there will be a major announcement on Twitter / Nvidia Monday morning at 9:00am.

http://twitter.com/NVIDIAGeForce

 

[dividebyzero]

@Per Hansson
My bad on the sound over HDMI/R600.

Having owned SLI'ed 8800 Ultra's and GTX280's (in one of my current rigs) I'll testify that multi-GPU drivers are not all sweet sailing-I think thats a given regardless of whether the setup is SLI or Crossfire- but have never encountered an instance where they did not work (the cards). While they may in some new profiles offer reduced performance and bugs these were usually rectified within a new driver release.
CD also asserts that the 8800GTX is some kind of white elephant...really, was it ?
Regarding bumpgate, while CD for the most part got the story right-and really this has been probably his only major journalistic coup- he still asserts that there is a 100% failure rate (which is what I was pointing out) which also includes G92 parts....100% ? Really ?

The point of my post was to outline that Mr demerjian's endeavours are usually couched in absolutes, which are seldom applicable in real world scenario's.
BTW:
[/off topic]
I would welcome viewers of this thread to post links quoting mr demerjian's critiques on:

AMD/ATI's Crossfire/CrossfireX drivers
The expose on how AMD sat on user complaints regarding the GSoD for 3+ months
RMA rates for factory overclocked HD5xxx series cards
X800XT PE retail availability after card launch
R600 v G80 performance revisited once both architectures launched.
Heat related failure rates of HD 4870 cards.
Driver profiles for the HD 4850X2
RMA rates for HD4870X2
With 40nm wafer yield problems solved HD5xxx series cards remain severely overpriced.
Why AMD with a successful CPU and GPU portfolio have not surpassed -in profit or brand awareness- a company rival that has neither a chipset business nor any new product line.
[/on topic]

Until Comical Charlie addresses the industry as a whole I would treat his "writings" as the slanted biased jottings that they are. Stating that the GTX480/470 are going to be sold in very small numbers is a no-brainer. Retail consumers aren't going to flock to these cards because of pricing and nVidia would make a much larger profit by diverting the most of the line to the professional Quadro and Tesla SKU's. If these cards were meant to be large volume then surely the HD5970 should also be viewed in the same manner.
One last point ; anyone with a reasonable grasp of probability and statistical mathematics might want to take a closer look at the numbers argument in the linked story.

EDIT: Mr demerjian is reporting that according to his "moles" the GTX 480:
-operates at a tempreture of 70C at idle
-has a fan speed of 70% at idle
-cannot do DirectX 11 - except in the Heaven benchmark where it is twice as fast as a HD 5870
- is within +/-5% of the performance of a HD 5870. Apparently this is "far too slow" according to the author
The disclaimer here is that some, if not all the numbers, have been fudged to protect his sources.

 

[Guest]

Here's an interesting article from Anandtech. If memory serves me, it has an excellent overview of the problems relating to the manufacture of large chips buried somewhere within. The article would seem to support conclusions from the link above.

http://www.anandtech.com/video/showdoc.aspx?i=3740

 

[Vrmithrax]

I can see why the first Guest was confused. "Fiscal year" refers to what year it is when your year ends, and can be pretty deceiving. In the case of nVidia, their financial budgeting year starts in February 2010, ends January 2011, and is referred to as "Fiscal 2011"

Gotta love the bizarre logic out there, eh?

 

[dividebyzero]

...In the case of nVidia, their financial budgeting year starts in February 2010, ends January 2011, and is referred to as "Fiscal 2011"

Gotta love the bizarre logic out there, eh?

It's catching !
How about a 2012 Ford Focus...just the thing for cruising during the apocalypse.

 

[timeless52]

So, what has happened to the Fermi chips? It certainly can't all be TSMC's fault; ATI seems to be getting an ample supply of 40nm chips.

No matter what you think of Charlie, his "speculations" here remain plausible. Only time will tell. Hopefully the answers will be coming forthwith.