AMD: R9xx Speculation

Seems like a waste and a good way to po some customers. On the bright side the lack of info has made it painfully obvious where everyone stands on their beliefs of what the card is. Those that think it will be a monster grasp at any info pointing to such, those that think it will be a dud do the same and anywhere and everywhere in between.

 

One of my great regrets is not keeping a log of all the tags the xtremesystems guys put on the Fermi thread. They were awesome.

This isn't one of them, but:

​

Damn, Cayman might be a bit crap but this is all hilarious.​

 

They'd probably end up equal after Anand "normalised" his fast board.

 

I was kind of joking, but if AMD did send slower samples, those would be development samples, not units sent to reviewers. That would obviously create a big mess.

 

Who's to say that wasn't a card sent to AIBs etc., that doesn't have to be for reviews

 

The card is so fast that Slovenia has been warped to Dec 12th.

 

I kept a log (also for a few other threads), I could post it if you want.

 

Thermonuclear Meltdown/Thermonuclear Letdown were mine.

 

I'd like to see Charlie acknowledge that, while it struggled out of the gates, Fermi has proved it's chops. No way he'll slam AMD the way he did nv for a flubbed release...

 

Hmm is it faster or is it slower, that's the question...

 

not being able to beat a chip that's 25% larger and uses nearly double the power = flubbed?

 

Really? I must be misreading Techreport ´s review of the GTX 570.

Dirt2 DX11, Metro 2033 and Lost Planet 2 shows GTX 570 to be 20% to 35% faster than 5870. Let alone GTX 580.

 

Neat, thanks, that would be cool - might be better as PM though.

 

Imho it's a reminder for when the NDA end

 

Sent, hope you enjoy.

 

Well, I'd say since

a. GT2xx and Evergreen are both the ultimate evolution of dated G80/R600 architecture.
b. AMD seems to be lagging on "generation" e.g. their K10h is almost 2gens behind (ix, ix2).


5870 should really be compared to GTX285.

 

You know what I find interesting? Less than 5 days to go and not a single die shot has been found. This must be some kind of record

edit: heck, not even a pcb..

 

Well, their goal was obviously to be on top and many of us thought they would nail it, but brute force won and NV deserves credit for it. That's all.

 

Says...?

 

I'm sure they'd love to have the highest performing single chip, but I'm not sure they would be willing to build something near as big as the 580 to do it.