AMD: R9xx Speculation

Who knows what the 6830 is, anyway? Maybe AMD won't cripple it as badly as the 5830.

 

http://we.pcinlife.com/thread-1500103-1-1.html

 

For reference, the HD 5870 gets 33.4 FPS in Crysis, apparently with the same settings: http://techpowerup.com/reviews/Zotac/GeForce_GTX_460_Amp_Edition/10.html So that's a rather massive 30% improvement.

 

Out of my memory, looks like it's getting about same FPS as GTX480 in Unigine @ Extreme?

 

It's 20% faster than GTX480 in Unigine.

http://www.hexus.net/content/item.php?item=25391&page=6

1,920x1,080 with 4x AA and 16x AF

 

That is a nice speed up in performance ..

I don't think it is attributed to a mere increase in SPs count though , something else must be at hands here , maybe the 4D shaders rumor is true after all ?

 

Uningine shows that tesselation performance are more than doubled!
With 8xAA, at extreme settings, the 5870 scores 304.. and without AA scores around 500.

 

It seems that they worked also on the tesselation unit.
I think that the most interesting part is yet to come: the die-size. What about it being smaller than Cypress?

 

The platforms are obviously different, and results with even close to same platform are hard to come by, but ~30%'ish in Crysis, ~30%'ish in 3DMark Vantage Performance, ~45%'ish 3DMark Vantage Extreme, ~75%'ish in Heaven Extreme
(I think)

 

Did they? The tesselation unit, as far as I've understood, never was the real problem, but the setup/rasterizer not being able to feed it fast enough?

 

Hah yes that would be too funny, but according to charlie in his forum...

http://www.semiaccurate.com/forums/showpost.php?p=63824&postcount=13

 

Nevermind.

 

Where's the link to charlie's quote?

 

Why does it say Vista on the Crysis benchmark but 7 on GPU-z? Is that normal?

 

http://www.semiaccurate.com/forums/showpost.php?p=63824&postcount=13

Don't know why the qoute did not work correctly, i simply copy/pasted it

 

With that die size, It doesn't really fit in the sweet-spot strategy.. maybe when they realized they couldn't use the 32nm process, they decide to simply "bloat" (an inverse die shrink?) the chip to 40nm, instead of reengineering the chip. So Cayman and all the other chips may have exactly the same specs as the canned 32nm versions. Reengineering all the chips may had have an higher cost..
Cayman@32nm would fit in 280 mm^2, not far from RV770 die-size, better size for the "sweet-spot strategy".

 

It's pretty impressive to see these kinds of gains. I'd really only expect that from the jump to 32/28 nm. It makes you wonder what AMD can do with the 28 nm jump next year.

The real question is if AMD has still managed to keep the size/heat/power/noise down. If so, it's going to be another year of AMD kicking Nvidia in the nuts, only this time 40 nm availability should be a lot better for AMD, so the kicking will be harder and more frequent.

I feel better about skipping the 5xxx series, and will be mighty tempted to upgrade to 6xxx at the end of the year, despite the fact it's still a 40 nm generation.

 

If Charlie is right and the die size is 115% Cypress for 130% overall performance then it looks like the architecture is a winner.

It does look like 40nm problems still persist somewhat because judging from the leaked code-names each performance tier (chip) seems to have three variants. I guess they are planning ahead this time, so there ought to not be any red headed step children in the lineup like the 5830 was.

So if the above holds true, I wonder what kind of price scaling we're looking at?

$399for the 6870 with 2GB Ram?

$299 for the 6850

$249 for the 6830

$199 for the 6770

$149 for the 6750

???

 

I m afraid it will be higher. If Cayman is bigger than Cypress, it means that less Cayman chips will be produced per-wafer, which means less profit for ATI. The GTX480 is the fastest card around and it's sold for 499$: if Cayman is faster, i would expect the same price. So at least 449$-499$. I guess the good old times of RV770 are over.

 

I originally wrote that, but then I thought "No, Dave wouldn't be THAT cruel."... In any case I don't think its the wafer price / yields which will dominate the price, I suspect it'll be the RAM for the highest end SKU and the fact that they don't have any competition at that price bracket. I guess introducing it at a higher price would mean they could clear out the older Evergreen GPUs without having to slash the prices too badly.

I guess they are going to be using some of the new 2gbit GDDR5 chips?