While it's nice to see a new bit of silicon it's a bit disappointing that it was a sideways move. A little bit more efficient, a little bit better performing. All kind of lack lustre I feel. Well, it's time to start picking up r9-280's though.

Also, any news on the r9-285X?

Too bad for AMD that the new Nvidia 960 is only 6-8wks away, which is likely to have performance 30-50% higher than the current 760.

..back to square one for you AMD.

Bagpuss
Too bad for AMD that the new Nvidia 960 is only 6-8wks away, which is likely to have performance 30-50% higher than the current 760.

..back to square one for you AMD.

Nvidia 960? or 860? #just asking

lumireleon
Nvidia 960? or 860? #just asking

It will be the GTX 960 as the 8XX series is mobile chips only. So people don't get too confused with naming structure. Though, I've seen no mention of a GTX 960 being launched any time soon. And Bagpuss is at best guessing at the performance, unless there's an article referencing a rumor or potential spec?

Hexus, your table on page 1 has a typo:
http://hexus.net/tech/reviews/graphics/74033-sapphire-radeon-r9-285-dual-x-oc-28nm-tonga/

The 280 and 280X are still GCN1.0, same as the 7970/7950.

edit: I think it's a quite promising chip - that 4k does not suffer in the slightest from the framebuffer is amazing, and this card (unlike the 280 it's replacing) supports FreeSync, which is possibly one of the bugger improvements in gaming quality of recent times apparently.

Are those die characteristics correct? Seems weird to have a larger chip with more xtors and yet only a 256-bit bus on it. If that's right then I'd have to assume the faster chip has the same 384-bit bus as Tahiti.

Jimbo75
Are those die characteristics correct? Seems weird to have a larger chip with more xtors and yet only a 256-bit bus on it. If that's right then I'd have to assume the faster chip has the same 384-bit bus as Tahiti.

It's cheaper to manufacture than a 384-bit bus.

I believe AMD is holding something back for the R9 285X. It does seem strange that the die, if it is a full implementation, isn't smaller.

Tarinder
I believe AMD is holding something back for the R9 285X. It does seem strange that the die, if it is a full implementation, isn't smaller.

It does pack extra features however, TrueAudio, the new UVD decoder and increased tesselation capabilities certainly have to take some die space too. And there might be many minor internal architectural differences, like larger caches and whatnot that also need some die space.

gonchuki
It does pack extra features however, TrueAudio, the new UVD decoder and increased tesselation capabilities certainly have to take some die space too. And there might be many minor internal architectural differences, like larger caches and whatnot that also need some die space.

While true, the bus width would still likely take up far more space than all of that.

Not only that, but AMD has already got the R9 290/290X bus down to a (comparatively) much smaller size, and that's 512-bit.

They're definitely up to something here, those numbers just don't add up to anything near what I would expect given the maturity of the process. Don't be shocked to see this is actually a 512-bit chip.