I have not read or seen anything that shows the hd7970 Tahiti GPU to have any parts of it disabled. Also look back at my previous comment I edited it and asked you question regarding Kepler.

Hasn't Nvidia already said Q1 for the 600 series?

You read it here first. Is it not logical that AMD would take a page from their old playbook?

And it will also be in MY own preview tonight. i just have to formulate it better there than i do here.
-- i have heard this from my own sources; it is not confirmed and probably won't be until the refresh in a few months.

And i don't "know" what Nvidia will do other than by their history
- if AMD launches their flagship card, Nvidia will respond with a faster one (if possible)

Nvidia will not meet AMD with their midrange. Unless the GTX 660 Ti is faster than the HD 7970, and then maybe they would do it as a "first" just to embarrass their competitor. (i would in that case, only)

i don't think any site makes this clear

The launch of the HD 7970 is Thursday, Dec 22 - the launch of the HD 7950 is on January 9th
- that is the day that you can actually buy either one.

So you are saying the full Tahiti GPU has either more than 2048 cores and/or a bigger than 384-bit bus?

i'm saying that some of Tahiti's parts are likely disabled to rush out this part ahead of Nvidia and to maximize poor yields.

AMD is not going to advertise this any more than they said that x1900xt was coming a few months after x1800xt arrived to stop the bleeding back then.


Does the GTX 580 have a wider bus than the GTX 480? No
- more cores, of course


AMD is very likely doing what Nvidia did recently with the GF100. The core is clocked low is just one clue - 925MHz with room for overclocking over 1GHz. That lets the partners come up with all sorts of cooling and clocking designs - in perfect timing for CES (ahead of Nvidia with their "old" GTX 580).

The next iteration of Tahiti chips will likely routinely hit well over 1GHz with all parts enabled and with lower TDP - just like GF110 did when compared with the GF100. Tahiti 'Next' is what AMD is putting up against the GTX 680 in Q2. Or they will be slaughtered performance-wise. However, i don't think this will endear them their fans.

Hmmm interesting. I'm certainly not one of the few who are in the know with what exactly is going on, and not that I am doubting your logic or sensibilities, but I don't think there are parts of Tahiti that are disabled that will affect gaming performance. I think this might be a case of Bulldozer - it can overclock like crazy but they weren't able to hit their target core mhz frequency within a certain TDP. So while there aren't parts of the chip disabled, they are going to do a deep respin at some point to release hd7980 or whatever they'll want to call it. It will be functionally the same, but operate with a core frequency of 150mhz and have the same or lower TDP. 1024 cores is a nice even number and there has otherwise been zero indication in the leaks that any parts of the chip is disabled. So yeah, I think they're going to rely on a respin to save face, but nothing other than a lower TDP and higher core operating frequency will be achieved. This is their first full on multipurpose GPGPU from AMD - it was a necessary step they had to take to stay competitive and to shrink down the number of different designs they would be working on at the same time. That it isn't going to be a crazy performance monster isn't very surprising to me.

Whether parts are disabled or not ...
(and the relatively low clockspeed, high transistor count, and comparatively weak performance indicate it is crippled)
.... The HD 7970 is simply not enough to meet the GTX 680.

Logic says so. That means that AMD released this chip as a stop gap. Their timing is very good. (1) To gain the single-GPU performance crown, to (2) give their partners new designs at CES and to (3) raise some quick cash from the adopters who must have 'the fastest'.

Smart .. in the short run .. we will have to see how their new CEO's strategy plays out

Where does this image come from? it is not an AMD slide


And in this one with 32 ROPs, Embargo is misspelled. AMD does make mistakes, but the specs may be deliberately off here. i need to check with my contacts


i don't know for sure if the chip is crippled or not - like GF100 was. But the low clockspeed, large die size and poor performance indicate it might be
http://www.chiphell.com/thread-329802-1-1.html
More or less translation: "Beats GTX580 by 20% on average, not 3DMark. Beats GTX580 by around 30% in Crysis. Expected goal has been reached. There has been no goal of doubling the performance of previous generation cards as speculated."

i can't believe it took until 3:30 AM to write it

NO CAFFEINE

http://alienbabeltech.com/main/amds-upc ... card/all/1

i hope you find it interesting and it proves to be mostly factual

Interesting. i just *now* got my first update from Edelman (AMD's new PR)
- from the CPU division