They did release the NV 30(Aka 5800 ultra),and like Fermi,it was late,used more power,cooler was noisy as hell and a lot more expensive than the 9700 pro...
About the only thing left not to make it NV 30 revision 2.0 is how it performs,but Nvidia is still keeping that a secret so it's not exactly confidence inspiring here,though i'm hoping for the best,since with both companies competing for our money,it means faster cards,sooner,for less money.
Considering the amount of holdup's they've had so far,in my view,the first generation of Fermi gives the impression that it simply isn't that great to begin with...Patience and understanding only go so far,and we're not that far away from the 5 month mark since the HD5870 was released.
so let's hope that this thing really gets released in march and that availability is actually decent,and not just slightly better than a paper launch where a lot of users won't be able to find one,even after waiting for them this long.
In short,HD5970's are monsters,and ATI are going to EOL the current HD5800/5900 lineup by the end of march supposedly,as new refresh cards are going to be released by early april it seems.....Right when Fermi is supposed to be released incidently.
Looks like ATI are already preparing a warm welcome for Fermi.
Anyhow,the overall situation for Nvidia is pretty crappy given that ATI has basically introduced their entire lineup of DX11 cards,covering from 60$ budget cards,all the way to the highest end ones,before nvidia releases even their first high end DX11 model,so they're lagging behind in a huge way here,not just on the high end,but all price points and models,which the lower end versions are the ones that sell in large volumes and pay the bills.
Anyone care to bet when we'll see a midrange 250$ card based on the fermi architecture from Nvidia?
It's really a bit of a strech to state it's the supposed release period either,just unconfirmed rumors,even if they're coming from several hardware sites....The definitive answer is with Nvidia,and they've been silent on release dates,performance,power use and prices.
With ATI possibly releasing refreshes of all their cards in that same rumored timeframe that Fermi is supposed to be released,and even if Fermi did have a performance lead with the original HD5800 lineup,will that be the same with the refresh versions?
Has fermi's window of oportunity to shine as the fastest card on the market come and gone,given all the delays it's suffered so far?.....That's the main question.
If it's over 500$ and finally gets released,beating the current HD5870 cards by the rumored 20%,then you'll have the same situation that existed between the HD4870 cards and the GTX280,or between the HD4890 and the GTX285 cards.....Nvidia charging a large premium for a relatively small lead in overall performance,even if you discount the card being released 6 months late,compared to when the HD5870 was released.
And it still doesn't beat the HD5970 cards anyhow,not to mention that there could be refresh cards from ATI shortly after Fermi is released,so Nvidia really screwed up here,at least when it comes to the first generation of the GF100,and it's time to look past that to be honest,into the second or third generation of Fermi,hopefully using a smaller fab process.
Ouch if true,and even though it comes from charlie over at semi accurate,which is widely known he's got no love for Nvidia to say the least,he has been pretty accurate with regards to fermi issues over past 6 months,so his sources,whoever they are,have been providing accurate information.
Seems the A3 yeilds are still way too low for any sort of high volume availability,and that to fix it and the high power draw,doesn't require a simpler and faster metal respin like what they've been doing for the last 3 revisions so far(A1 to A3),but a full base layer respin which would include new masks,and delay fermi at least another 3~6 months in the process.
Also seems that the clock speeds for it hit 600 Mhz at most,even with 64 SP's disabled,and only 5000 to 8000 cards available on the first run,with the actual performance lead with the reduced clocks and shaders will be about 12% faster than cypress overall,and that's with the card using 280 watts of power,versus the 188 watts than cypress uses,while the HD5970 remains untouchable basically....This is not good