Nvidia Geforce GTX480 "Fermi" is broken and unfixable

jmke · 17th February 2010, 21:46

We first got word that production A3 GF100s were back in Santa Clara near the end of January. When enough of them rolled off the line to characterize the silicon, we hear there were no parties in Santa Clara. For those not interested in the 'why' side of things, the short answer is that the top bin as it stands now is about 600MHz for the half hot clock, and 1200MHz for the hot clock, and the initial top part will have 448 shaders. On top of that, the fab wafer yields are still in single digit percentages.

That said, the situation is far more nuanced than those three numbers suggest, and the atrocious yields are even after the chip has been downclocked and defective units fused off. To make matters even worse, the problems that caused these low yields are likely unfixable without a complete re-layout.

http://www.semiaccurate.com/2010/02/...and-unfixable/

leeghoofd · 17th February 2010, 21:49

John semiaccurate is the new fuadzilla... if he can bash Nvidia he will, being true or not, people still readup on his articles... amasing...

Take with a big grain of salt and sometimes better a whole package

17th February 2010, 23:48

Decrease the severity of anything Charlie says about NVIDIA by half and cut out anything clearly intended to raise the ire of fanboys and you've got something relatively close to the truth, usually.

18th February 2010, 01:11

I think the info makes far more sense when combined with what Anand has said regarding TSMC's problematic 40nm process... Really bad timing to be launching a huge chip on that, even ATI needed a few extra tries before they worked around the TSMC process problem.

18th February 2010, 05:53

Rather than read an excellent example of yet another long, biased tirade... try this article:

http://www.semiconductor.net/article...fects-full.php

jmke · 18th February 2010, 16:07

Hey Albrecht, I'm well aware that SA has to be taken with a huge grain of salt, his articles are provocative, not always based on fact, but derived from rumors and guesses;
nevertheless it's interesting to follow his train of though and see what conclusions he draws from the data he somehow obtained;

so far, the accuracy of reports by TheInq,Fudz , SA and other rumor sites has been increasing over the years, where as before any article posted by TheInq was completely fake and debunked when the actual product was launched; but nowadays, the guess work seems to be a lot more in the ballpark, sometimes right on the dot.

all things said, those sites are fun appetizers while we wait for main dish

next article:
http://www.semiaccurate.com/2010/02/...t-want-answer/

18th February 2010, 18:53

Well I read his articles because he'd tell you facts, then spin them his own way... could at least get some facts out of his articles. But it gets annoying to shift through blatantly made up spin, it's worse than the news media. He even links to Anand's article, yet his goes against half of what was said in it. Lately he seems to be on hyper-mode with the hate... maybe when he writes less like a angsty 18 year old I'll go back to reading that site.

jmke · 18th February 2010, 19:13

Quote:

Reply from nvidia.
http://twitter.com/RS_Borsti
Oh Charlie... that just another hilarious post

http://twitter.com/Igor_Stanek
Oh Charlie... that just another hilarious post me: I think with this post Charlie totally destroyed his credibility

src: XS

17th February 2010, 21:46	#1
jmke Madshrimp Join Date: May 2002 Location: 7090/Belgium Posts: 79,022	Nvidia Geforce GTX480 "Fermi" is broken and unfixable We first got word that production A3 GF100s were back in Santa Clara near the end of January. When enough of them rolled off the line to characterize the silicon, we hear there were no parties in Santa Clara. For those not interested in the 'why' side of things, the short answer is that the top bin as it stands now is about 600MHz for the half hot clock, and 1200MHz for the hot clock, and the initial top part will have 448 shaders. On top of that, the fab wafer yields are still in single digit percentages. That said, the situation is far more nuanced than those three numbers suggest, and the atrocious yields are even after the chip has been downclocked and defective units fused off. To make matters even worse, the problems that caused these low yields are likely unfixable without a complete re-layout. http://www.semiaccurate.com/2010/02/...and-unfixable/ __________________

17th February 2010, 21:49	#2
leeghoofd [M] Reviewer Join Date: Dec 2008 Posts: 3,209	John semiaccurate is the new fuadzilla... if he can bash Nvidia he will, being true or not, people still readup on his articles... amasing... Take with a big grain of salt and sometimes better a whole package __________________

18th February 2010, 16:07	#6
jmke Madshrimp Join Date: May 2002 Location: 7090/Belgium Posts: 79,022	Hey Albrecht, I'm well aware that SA has to be taken with a huge grain of salt, his articles are provocative, not always based on fact, but derived from rumors and guesses; nevertheless it's interesting to follow his train of though and see what conclusions he draws from the data he somehow obtained; so far, the accuracy of reports by TheInq,Fudz , SA and other rumor sites has been increasing over the years, where as before any article posted by TheInq was completely fake and debunked when the actual product was launched; but nowadays, the guess work seems to be a lot more in the ballpark, sometimes right on the dot. all things said, those sites are fun appetizers while we wait for main dish next article: http://www.semiaccurate.com/2010/02/...t-want-answer/ __________________

17th February 2010, 23:48	#3
Oberon Posts: n/a	Decrease the severity of anything Charlie says about NVIDIA by half and cut out anything clearly intended to raise the ire of fanboys and you've got something relatively close to the truth, usually.

18th February 2010, 01:11	#4
Kougar Posts: n/a	I think the info makes far more sense when combined with what Anand has said regarding TSMC's problematic 40nm process... Really bad timing to be launching a huge chip on that, even ATI needed a few extra tries before they worked around the TSMC process problem.

18th February 2010, 05:53	#5
Kougar Posts: n/a	Rather than read an excellent example of yet another long, biased tirade... try this article: http://www.semiconductor.net/article...fects-full.php

18th February 2010, 18:53	#7
Kougar Posts: n/a	Well I read his articles because he'd tell you facts, then spin them his own way... could at least get some facts out of his articles. But it gets annoying to shift through blatantly made up spin, it's worse than the news media. He even links to Anand's article, yet his goes against half of what was said in it. Lately he seems to be on hyper-mode with the hate... maybe when he writes less like a angsty 18 year old I'll go back to reading that site.