Yeah, Charlie knows his stuff all right. I don't know why people despise him so much for it and accuse him of talking hot air. He's cutting through the corporate spin and crap and reporting the truth to us all. What's not to like? Comments complementing Charlie always seem to attract flames unfortunately... please keep it civil everyone.
Because he is not right at all. He is semiaccurate at best and that works both ways, he can be right or not and most of the times he is not. Now he bends the truth to look like he was right, but he was not right and beither is now probably. i.e:
- He says less than 10.000 cards are going to be launched, despite Nvidia disabled a SM. Several other sources with far more reputation than him are saying that Nvidia had less than 10.000 only when it was 512 SPs, now that they disabled an SM they will have 50.000 cards at launch. Charlie 0 - Truth 1
- He said Fermi die was greater than 600 mm^2, it's 530mm^2. It's easy to say the chip it's going to be big, everybody knew that. That's the same that clairvoyants do, use vague words, that way you are always either semiaccurate if you are right or only semi-inaccurate if you are wrong, pretty much like: a very famous and loved celebrity is going to die this year. No man, you have to be 100% right on this one or you are NOT right. He was right? Definately not. Charlie 0 - Truth 2
- He said it would NOT have hardware dedicated tesselator and made many claims, articles and long posts and thorought explanations of how and why, both in SA and Beyond3D. (In fact he also said it would not have any dedicated hardware like ROPs or TMUs, but let's forgive him since that was long ago). Charlie 0 - Truth 3
- He said high end Fermi had been cut down to 448 SPs. Again or you are right or you aren't. 0 - 4
- He said the clocks would be 600- 625 Mhz for the high end. That was last week, specs were final by then, if he trully had any info he would have been right. 0 - 5
- He said that Fermi would run at 1.1 volts, now he says that Nvidia actually upped the volts, even when the card runs at a much lower voltage than what he claimed. Really? 0 - 6
- he said the 512 SP part was only 5% faster than 5870 (he still says that, from what I read) and the 480 SP part is actually 15-20% faster, while the 448 SP 600 Mhz part is almost as fast. 0 - 7 (In fact, not so long ago and for a very long period of time he said that Fermi would be much SLOWER than Cypress actually, so I'm really thinking about giving him 2 negative points here instead of one)
- He said 2% yields, because they gave him this info: "Nvidia got 9 samples back from TSMC in September" when the message, before it was badly translated from chinese actually said: "Nvidia got samples back from TSMC in September 9th"
0 - 8
- He still pretends that Nvidia asked 9000 waffers for Fermi and will release less than 10.000 cards, so he pretends to say that Nvidia is selling cards at a loss of $3000-$4000 per card.
0 - 9
- he said it would be late. 1 - 9
The only thing he is been right about is that it would be late so far. That and that it would have a very high TDP, but that's something that he didn't say himself, I mean he was not the first one to say that. I think it was brightsideofnews.com. we give him one point? Hmmok... 2 - 8
All in all he has never been right and he said the same things about G80 for example. In the case of G80 everything was wrong or semi-wrong and in this case things have been semi-accurate, if you take the vague claims instead of the actual "facts". That is, it's late (although not as late as he claimed, April anyone?), it's not as fast as Nvidia wanted although it's not as dramatic as he said.