Yes, your revisions are off by one (they start with A1, ATI starts with A11)
According to the partner slides, they were hoping on shipping the first samples to their partners in Late October, enabling a late December launch. This would've been done with an A2 revision on risk wafers.
But a shipping A1 was never viable, GT218/6/5 all had one spin (in February) before going into production. After the Cypress launch it became clear that the most optimistic date AMD set for a Fermi launch (and which nvidia communicated to it's partners) would be February. CES will be a sneak peak of Fermi much like AMD did with Juniper/Broadway at Computex.
After A3 went into production (beginning of December) reported yields were still low, coupled with TSMC's general problems (capacity being the first) saw them slipping the launch date to somewhere after CeBit, this was communicated just before newyear in a "waiting for enough parts
If you take the Fudzilla articles, they hold a lot of truth, dual-Fermi, top-to-bottom stuff, but you have to be really selective to see the facts in there and distinguish them from the marketing. Original plan was to launch GTX395 almost a full quarter after Fermi, and the mid-range parts in the quarter after that. Now everything seems to be going really well on mid-range and low-end Fermi parts so we might see a similar launch curve like AMD has done, in 4 months time launch everything from high-end discrete to low-end mobile.
P.S.: I'm sorry for this shallow and extremely biased first post. Old fanboys never die (they just buy the competing hardware in the end.)
P.P.S.: Yes, Fermi mid/low/mobile will be (a lot) better than the current GT21X's
P.P.P.S.: No, I'm not Charlie's Stepson/Husband/Lover/Room mate.






Bookmarks