Yeah, i noticed that TheHunter didnt get it. I hope that they release this shrink/new card/whatever soon, it would be cool to see 28nm in action.
I'd say this is the beggining of the 5xxx refresh IMO. Imagine 2x5770 256bit gddr5 28nm cfx 1.4ghz each card with oc lol that'd be nuts
Good day guys, , will it could be more advanced chips but, , curious the 32 nm was hard enuf to master the silicon compound wouldnt handle the preasure of the transistors 28 nm, even more impressed, , anything is posible , the 2nd week in december intel lost the crosslicence for amd64, , / x64 , and for there own licence amd sued for 1 250 000 000 dolars, end of december the money was in, , but it was a 18 month court case, ull find that google , finance that amd equity grew with 1.25 bil $, that chip could be leo , the new amd platform, or a better die for the hd5 series,
+1 ATI doesn't normally start the next gen on a new fab process. Needless to say, if this is a 28nm GPU, I'm almost certain it's a refresh of the 5xxx series. Encouraging to see Q1 2010 as the release date for 28nm GPUs- just about the time I plan on upgrading too. And 384bit bus width? Niiiiice. Unless the rumors of Fermi being ridiculously hot, power hungry, and only 20-30% faster than Cypress are completely false, nvidia is in serious trouble... Thanks for the link, btw.
*Chubu (who is atheist) starts to pray to God so that the 5790 launches with a price (and availability) that he can afford.
From a business perspective, AMD would be smart if they started with their refresh/6000 series now, with the shortage of 40 nm chips in mind ( when both companys where looking for silicon at the same time frame ). If AMD can beat nvidia on the 28 nm silicon and drain the manufacturers now they will be in an unbeatable lead position once it is time for a die refresh or new series. I do not think Nvidia has the founds to look further then 40 nm at the moment. They are far from broke but they didn't make as much money as AMD did %age wise, and their potential earnings from this serie is alot lower since they are 6 months late to the party. Their potential buyers are people that: 1) W8ing exclusivly for Nvidia GPUs 2) People w8ing for a pricewar, going with whatever gives most bang / buck and by history AMD usually delivers on that area, having 6 months headstart doesn't hurt AMD in that aspect either since they will have gone -+ 0 by now. 3) People on the fence, not sure if they should w8 for 6000 series or buy 5000/Gt300 series ( most likely people sitting on 285 sli, 4870 Tri fire, quad 295s ) with no real viable upgrade path and no real "epic" dx11 titles that are must haves released YET. Without knowing jack about what goes around at nvidia or AMD it would seem like AMD would have the most incentive to be early on 28 nm this round; they have a good market position now and the ball is in their corner so to speak. Nvidia cannot "afford" not releasing any 40 nm dx11 and they can litterally not afford going too early on 28 nm since they havn't made any earnings from 40 nm so far but invested alot in development of the new chip. If I where AMD I would be early on 28 nm to further preassure nvidia, and with the 40 nm silicon shortage in mind beeing early means beeing the first to make money of the 28 nm ( since it isn't unlikly that 28 nm will be in short supplies at first as 40 nm where ). Interesting times for sure. Oughta be good for us consumers as long Nvidia have the marginals to go into price war
AMD already have their 6000 series road plan mapped. It's a new architecture so it would be unwise for them just to all guns blazing. They have all the usual processes to go through like test silicon (which is what this wafer could possibly be) and then their inital test boards. AMD are said to be skipping the 32nm node and going straight to 28nm so its totally acceptable that this wafer is at least test silicon to get an idea of the problems they may face when they start fabbing their engineering samples.
so this will be ATI's answer to Fermi? I LIKE! By the time Fermi is out everyone will say wohooo nVidia took the crown and 2 weeks later... a 28 nm based graphic card from ATI comes along and takes the crown again:banana:
Yah ofc, they won't roll out something called 6870 during 2010, but they will start on it and as with the 4000s it will most likly be a refresh as with 4890 wich took the 4870 from 750 to 1000 core and added like 15-18% performance. But that aside I think ( without knowing ) only AMD would be ready to order / look into 28 nm at this stage. Really hope fermi isn't over hyped and that it delivers, else it is gonna be bad for the prices.
That wouldn't be a particularly wise choice for a cash-poor AMD. Especially considering that they could easily close any performance gap with Fermi by releasing a 28nm version of the 5870 and clocking it higher and widening the memory bus. Every time you drop down to a smaller die, problems arise. Every time you tape out a new architecture, there are usually problems as well. The trick is to anticipate these challenges so that you never have to experience them simultaneously, as it may overwhelm the engineering team. Nvidia learned this lesson harshly when they developed the NV30 (GeforceFX) architecture. It was a new architecture AND had used a newly shrunken die process- 130nm, compared to Radeon 9700's 150nm process. Unfortunately for nvidia, there were yield problems AND architectural issues. The chip was delayed for several months, and the final product, GeforceFX 5800 Ultra, ran hot and loud and still couldn't compete performance-wise and cost-wise with the Radeon 9700 Pro, despite the fact that nvidia was selling them below cost. It was a disaster. Now tell me, why should AMD hurry the 6000 series on a brand-new 28nm process when they already have the performance lead? Which makes better business sense?
