Its very hard to know where glofo is with 28nm, people keep seeing varying numbers for dates on it being ready largely because they are talking about different fabs.
2012 dates for 28nm are for the New York State fab, there they'll be looking at well, huge capacity. AFAIK they have 28nm designs taped or taping out now, but their 28nm capacity would likely be very small scale, probably part of the Dresden fab but maybe in their singapore fabs, its unclear.
The conference in June will be on a tonne of things, though a lot of it is gpgpu/apu type intergration but also on where they want graphics to be heading.
AMD have, to date, rarely outted new tech that wasn't very very close to being ready.
Its more than possible that AMD and even Nvidia have non high end parts on 28nm due mid year, tester parts as opposed to full blown parts. As for when the high end stuff is due, no one really knows, taping out is one thing, needing respins can push any theoretical dates 2-3 months back. Historically Nvidia tends to have the most trouble delivering new parts on new processes quickly so we'll have to see how the next 6 months plays out.
How Glofo's 28nm plays out will be interesting aswell, keep in mind stuff like low end AMD fusion stuff getting new parts at 40nm, I'd imagine a LOT of their 28nm capacity from next year will go to ARM and Fusion products, which will tend to be made on mature "last gen" processes as opposed to jumping to buggy new processes instantly.
It could be Glofo keeps massive capacity in New York while they do small scale "new" processes elsewhere. Thats what TSMC do aswell, they started off at 9k wafers a month for 40nm, which then expanded to 25k or so almost a year later. GloFo could be bang on for 28nm, ahead of TSMC, or way behind, its not hugely well known.
Anyway, the annoucements are set to be GPGPU focused and potentially VERY big, as in they appear to be ready to talk about a realistic method for code to utilise both cpu and gpu at the same time with ways to determine what parts of code would be done best where, which is basically the next big step in making APU's(but also gpgpu) useful, and pushing gaming forwards. Lots of physics stuff is MUCH faster on a CPU, while a lot is much faster on a gpu, being able to use both, effectively, with no penalty would be pretty huge.
, me thinks sandybridge-E & new ati card should do the trick.
