GF100 (Fermi) previews and discussion

Page 9 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.

to be honest I wouldnt know msaa from msds,,,,,I know what msds is but msaa to me is about as useful a terminology tool as a material safety data sheet for antacids.
 
Some can and will wait, some wont. Some insist on waiting, ignoring better solutions at the time.
Im trying to point out, theres alternatives outthere, theres also alternatives coming, so, how long do we wait?
Im also trying to point out, that everything Ive heard is, Fermi will come in with low numbers at launch, and that a decent availability wont happen for awhile.
This could be wrong, but the same sources saying this are the same ones that said it wouldnt be here in sep, oct, nov, by Christmas, Jan or Feb, but was saying March all along.
Ive seen people waiting since Oct, thinking itd be here. Ive seen others that quit waiting and are happy with their alternative purchases. The option to wait is always there, and by that time, some else is always there as well, thats better also.
Thats not gloom and doom, thats how it works
 

her scar tissue not mine, I know its your mum but whaddever . It works like mad.
 
Here is a bit of info that Nvidea fails to tell us about; they are maxed on power consumption and die size, and the GF100 is comparable to the HD5970, and yet ATI has more than 50% headroom to improve before they max out, while Nvidea is against the wall. Also the chip yeild is expected to be around 20% which translates to a huge price tag with 0 availability at 4 times the price min of ATI's 5970 with minimal gains. Not to mention the ungodly temps the Nvidea card is going to bring to your case. Finally the synthetic benchmarks that Nvidea used are not real world apps, and a little skewed, and to top off the spin parade they put on, they compared it to the 5870 rather than the 5970 it is to compete against, why? you ask, because the performance against a 5970 would be an ugly comparison, because there are no gains at 4 times the price and no headroom for improvement. Also to be noted is that the card that was benchmarked was not the card that will go into production, which will be purposely handicapped to meet PCIe limitations. Here is a good article to read for more info: http://www.semiaccurate.com/2010/01/17/nvidia-gf100-takes-280w-and-unmanufacturable/
 


Describing Charlie's articles as truth is only reserved for fanboys. Those numbers are really exaggerated. And before you start calling me an nVidia fanboy please check my system.
 

I think it goes without saying that a new large die generally is yield impeded at first, what complicates it is nvidias 3 billion transistors on an immature node, things will get better over time, I am under the understanding that nvidia uses the same M arch for all cards some parts disabled or whatever on non workstation cards different subsystem chips and whatnot. It would seem apparent that nvidia needs to reassess their design IF ts true that they are against the wattage limits.

Charlie also claimed he coined the term k10 , they guy is a running joke in some circles;I definitely dont take what he says seriously, I cant bring myself to do it. Id rely on Fuad at fudzilla before charlie and fuad can get excited about rumors, he just doesnt treat rumors like the mud that charlie does and then follow charlie in rolling around in said mud like a pig.
 

Has anyone double checked his pricing and yield numbers? we dont really know how tsmc is doing on that chip do we?
 



LOL point taken, and yes some of his statements are pure speculation and trash, but the fact is that Nvidea has been promising this card since Q3 last year and has had nothing but problems, and the release dates are nowhere near set in stone for March. What it boils down to is that Nvidea may be looking at a major catastrophe if they can't get the cost down and the production up, and perform with significant gains over the 5970.
 

I dont think its likely that they wont be able to come back. They have the best workstation cards in the world and are brilliant designers. They have hit a snag but theyll get it worked out; It is interesting how ATI have put them in such a corner but to be realistic its not like they need desktop cards in any short order, ATI still cannot do the computation Nvidia can and that type of card in science and movie making etc etc is what really drives the market like servers vs average pc users.

They have all the time they need and the 295 gtx isnt totally antiquated just yet.
 
mp7570.... do you honestly believe half the crap that you type....... wait, nevermind you listen to Charlie for news apparently. He may have had some correct points over the past few months, but never anything that wasn't already guessed at and his track record is still horrendous.

No one has any info on this card yet, yet everyone loves to speculate on it.... if Fermi fails, call it a fail. Until then call it a red herring, because that's all we know right now is nVidia, Semi-Accurate and ATI FUD.
 
He had enough info to tell us not to expect it til as late as march or later, way before anyone else.
As for the layout of the arch, it appears it may be down scalable, not sure yet, but they really need it.
At this point, speculations running they run quad tesselation, 1 per "core", with full chip having 4, not sure about scalability doing this, time will tell.
ATIs approach is same tesselation HW, just less units driving it.

As for FUaD, he was still buying the nVidia PR, and claiming itd be here in Oct, then Nov, then by Christmas etc, all the while, Charlie was telling everyone, FUad was buying into nVidia PR...so, who ya gonna believe?
Charlies been much more right than wrong, and as far as nVidia goes this arch, hes been as right as others whove proven in the past on their early ideas all things ATI, which were also right again this go round on the 5 series.
If youve read him from the beginning, youd see, tho he hits hard against the green team, what he said this time was as good as specultaion as theres been on anything at any time.
I still approach everything with some salt, but you wont see others doing this, and also being so accurate, lil less semi, more accurate letely
 


Does saving someone's life redeem one for a lifetime of murder?

Until Charlie can be "more accurate letely" for a good couple years, I won't give him credit. Until that point, he, AND Fuad, are both horrible at reporting news.
 
i guess ATI is coming out with a new card in h2, seems they plan on pressing nvidia as much as they possibly can.

http://www.nordichardware.com/news,10630.html

AMD is most likely very happy with the new Evergreen family but already during the second of this year we will see a new generation of graphics cards from ATI/AMD. AMD VP Dirk Meyer revealed this during a conference talk with analysts where he spoke of an update of the company's graphics cards during the second half of 2010. Exactly when this will happen and what the new series will bring to top the current Radeon HD 5000 family is uncertain.

Back in October we reported on scattered details surrounding AMD's next generation graphics cards, but whether this is the new graphics card architecture or a refresh of Evergreen is still to be revealed.

We have been in contact with several of AMD's graphics card partners and so far no one has any distinct information on the next generation of graphics cards.


 
Ill say this for Charlie. This time, concerning Fermi, hes been better than anyone concerning its speculation. You want to go elsewheres with what hes said about other things, fine, but Im not, hes been pretty solid on Fermi, and no, you wont see those "reputable" sites getting onto speculation at all, which makes them pretty good as far as 20/20 hindsight goes, but thats a totally different matter, and I give those sites a fail, as they dont do it at all, and dont sticj their necks out, so its not even the same thing

You wont see Toms or Anands etc doing this, so apples and oranges
 
Some are impressed pretty easy. I'll make some predictions. Fermi will be faster and more expensive than Cypress. It will do DX11 with less performance hit than Cypress up and down the product line. Nvidia will not sell these cards at a loss and proof will be the value of Company stock in Q3.
I'm Physic !
 

there is a reason people dont stick their necks out its called credibility and it can make or break a company. charlie has little cred on the street.
 


So forgetting notty's condecending comments for a minute.

I don't believe Charlie has predicted anything really tough to figure out at all. Except for some very devout people (including Fuad in some cases), I don't believe there was anyone who "speculated" that the Fermi would be released in 2009. People hoped, people said it was possible, but no one said "it's likely".

Charlie just verbalized and wrote down what pretty much everyone assumed, it didn't NEED to be said. So you want to give Charlie credit for saying the obvious and then say it makes up for his epic past failures? I'll give Charlie credit for having the balls to keep pushing the stuff he does and call it news, kinda like Dominoes and Little Caesers calling their product "pizza". That is ALL he gets credit for.

There's a reason those "reputable" sites don't speculate. Because when you're wrong most of the time, people tend to think you are either bat-s-h-i-t insane, or that you're just plain ignorant. The smart thing to do, is wait and see what actually happens, and then report on it. On the other hand if you're usually right people think you're a genius; it's a bad gamble to take though, as history seems to prove over and over again.

I can speculate just as much as the next guy; if I want news I'll go to a news site, if I want speculation I'll ask my sister's dog.....
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom