AMD: R9xx Speculation

[Tchock]

Whoa, good pricing.

Struggling between $180 6850 and $260 470 (For Premiere Pro and FX apps)

 

[jimbo75]

http://www.hexus.net/content/item.php?item=27053

Hexus is live guys.

 

[SimBy]

Wait, is that nV slide?!

 

[thatdude90210]

In the mid-performance heap, the best way to get to the top is to claim it first.

 

[AlphaWolf]

Hardocp

 

[AwfuL]

ComputerBase:
http://www.computerbase.de/artikel/grafikkarten/2010/test-amd-radeon-hd-6870-und-hd-6850/

Hardware Canucks:
http://www.hardwarecanucks.com/foru.../37286-amd-radeon-hd-6870-hd-6850-review.html

LegitReviews:
http://www.legitreviews.com/article/1445/1/

TechPowerUP! (5 reviews actually.. one for every IHV. Never understood why they don't combine those results)
http://www.techpowerup.com/reviews/HIS/Radeon_HD_6870/

 

[Tchock]

http://www.computerbase.de/artikel/grafikkarten/2010/test-amd-radeon-hd-6870-und-hd-6850/

this one has the 6850/70 closer to the 5850/470... hm?

 

[SimBy]

http://www.hexus.net/content/item.php?item=27053&page=17

HD 6850 looks like a great buy if you overclock it to 900MHz.

 

[Dave Baumann]

Note on 40nm/32nm transition:

http://www.anandtech.com/show/3987/...renewing-competition-in-the-midrange-market/2

Ultimately 32nm was canceled around November of last year. But even before that AMD made the hard choice to take a hard turn to the left and move what would become Barts to 40nm.

Also note - Northern Islands was always Northern Islands, nothing else. The switch to 40nm resulted in a common abbreviation of "NI40", but they were always NI.

 

[DavidGraham]

http://www.anandtech.com/show/3987/...renewing-competition-in-the-midrange-market/2

However it’s worth noting that internally AMD was throwing around 2 designs for Barts: a 16 SIMD (1280 SP) 16 ROP design, and a 14 SIMD (1120 SP) 32 ROP design that they ultimately went with. The 14/32 design was faster, but only by 2%. This along with the ease of porting the design from Cypress made it the right choice for AMD, but it also means that Cypress/Barts is not exclusively bound on the shader/texture side or the ROP/raster side.

 

[ZerazaX]

Looks like theyre saving the savory details of the architecture for the Cayman launch...

 

[LonelyMind]

Congratulations, Dave, for some outstanding engineering and price points. These cards will be my easiest recommendations in it's class, by far.

 

[AlphaWolf]

Note on 40nm/32nm transition:

http://www.anandtech.com/show/3987/...renewing-competition-in-the-midrange-market/2



Also note - Northern Islands was always Northern Islands, nothing else. The switch to 40nm resulted in a common abbreviation of "NI40", but they were always NI.

Sure, now you tell us.

However it’s worth noting that internally AMD was throwing around 2 designs for Barts: a 16 SIMD (1280 SP) 16 ROP design, and a 14 SIMD (1120 SP) 32 ROP design that they ultimately went with. The 14/32 design was faster, but only by 2%. This along with the ease of porting the design from Cypress made it the right choice for AMD, but it also means that Cypress/Barts is not exclusively bound on the shader/texture side or the ROP/raster side.

So how many version of cayman are floating around?

 

[codedivine]

Nothing new for compute?

From a compute perspective, nothing new seems to be there and once again no FP64.

 

[Dave Baumann]

So how many version of cayman are floating around?

We look at many different potential configurtions. The two being discussed for Barts were effectively the same (target) cost so they were analysed more closely to establish which way to go.

 

[trinibwoy]

So implementing the tech in hardware, writing sample code, programming guides, tech demos, doing industry expo presentations & direct devrel for several hardware generations aren't enough?

And what was the result? You would think after 7 "generations" of tessellation hardware they would be blazing trails but instead they're stuck with complaining about nVidia as usual.

But of course there are subsets, lots of possible ones.
Yes, you either FULLY support DX11 spec or not. But you can have some tessellation feature in DX10.1 that are compatible with DX11, you just can't call them DX11.

Can you elaborate? I'm not aware of any DirectX interfaces for AMD's "old" tessellator.

Congratulations, Dave, for some outstanding engineering and price points. These cards will be my easiest recommendations in it's class, by far.

Yeah seconded. AMD is really keeping the ball rolling. Execution on the hardware side of things has been excellent the last few years.

 

[Kaotik]

Note on 40nm/32nm transition:

http://www.anandtech.com/show/3987/...renewing-competition-in-the-midrange-market/2



Also note - Northern Islands was always Northern Islands, nothing else. The switch to 40nm resulted in a common abbreviation of "NI40", but they were always NI.

One of your (AMD) reps said different though when asked, I don't have his name (yet anyway) sadly, but he said that NI became SI and later again NI

 

[DavidGraham]

Texture Filtering did truly improve and according to pcgameshardware :

1-HD 5000 filtering bugs were hardware in origin .
2-Filtering is fixed in HD 6000 series .
3-Default A.I setting (Quality) is actually worse than HD5000 .
4-High Quality AI improves filtering noticeably , and everything is good again.
5-Nvidia's High Quality filtering is still superior .

 

[Sontin]

Yeah seconded. AMD is really keeping the ball rolling. Execution on the hardware side of things has been excellent the last few years.

They dropped the ball with Barts. No real improvement with Tessellation. And they must decrease the default AF IQ to win over the competition and Cypress (sry, only available in german): http://www.pcgameshardware.de/aid,7...irectX-11-Generation/Grafikkarte/Test/?page=4

 

[Dave Baumann]

Can you elaborate? I'm not aware of any DirectX interfaces for AMD's "old" tessellator.

The Xenos tessellator is exposed via DirectX. There was a PC backdoor in the drivers for DX10 cards as well - there is tessellation whitepaper from Valve a few years back where they look at a few ways of tessellating and they compared using this as well.