AMD confirms R680 is two chips on one board

No. D9E is scheduled for March. R680 is scheduled for the last week of January.

 

Thanks - I did not knew that; I thought it was about same time frame with R680.

 

I'm not sure you'll find much in that vein if, as the rumors suggest, AMD's aiming for 15% faster. 15% won't mean the difference b/w playable and not. There might be the odd example like CoH DX9, but outliers like that tend to go both ways.

But I'm speaking from the skeptic's camp, as usual: less downside.

 

So we should expect both to be released on Valentine's Day?

 

Actually ATI R600Crossfire wins against single G80Ultra in more other Titles.
http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/page10.asp
http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/page12.asp
http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/page14.asp
http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/page16.asp
http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/page18.asp

Edit: If R680 only aimed 15% improvement over G80Ultra probably due to power-consumption and heat issues ATI has to resolve by lowering clock frequency for R680. (Same example as GF7950GX2 500MHz for each GPU compare to single GF7900GTX 650MHz GPU)

 

That's one benchmark though, there are others where it doesn't do as good:

http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/page20.asp

And then you have the RAM problem with dual-gpu solutions. And of course compatibility problems.

 

I will expect R680 will be slightly faster vs. 2X Radeon 3850, because short distance travel between 2 GPU's on one PCB running around 650MHz for each GPU for R680.

Example: Radeon 3850 crossfire
http://www.anandtech.com/showdoc.aspx?i=3151&p=10

 

Why do you think R680 will run on so low clocks?
Sample and clockrates were allready shown:

http://www.theinquirer.net/gb/inquirer/news/2007/11/17/amd-showcases-radeon-hd-3870-x2

 

Well the reasoning I though for heat issues like my Radeon 3870 @ 800MHz GPU gets up to ~86c with stock low RPM fan. I guess I was wrong about Dual RV670.

 

Core clock seems to be on the mark, but what's with that memory clock ?
That's way slower than HD3870... GDDR3 ?

 

Yes, maybe cost-saving to hit $399-499 MSRP target...

 

So does it mean it won't be equal to two HD3870's in Crossfire, but more like overclocked, 512MB HD3850's in a "Crossfire-on-a-stick" configuration ?

With HD3870 approaching sub-200 dollar levels, two of them might end up costing as much or even less than this one, while providing additional performance (due to extra memory bandwidth, especially important in high resolutions).

Doesn't seem like anything that would stir my interest, based on these preliminary observations...

 

PCB nudies of R680: http://www.techpowerup.com/

 

Interesting they're using a PCI-e switch between the two RV670s. So much for synergy between AMD & ATi. HT would've been the obvious choice there. I guess it was cheaper to out-source it.

 

I'm getting scary V5 6000 flashbacks over here, especially seeing that PCI-e switch.

@no-X: I'd say they're the same size due to how the ATI card is at a slight angle/bent at the back relative to the camera.

 

It's aligned in compliance with the bottom margin - capturing angle shouldn't affect accuracy of the comparision.

 

See how it bends away from the camera/white lines. If someone was really really bored they could do some maths to figure out the loss in the perceived length and work out its actual length. I've seen people count pins and analyze the res of Halo 3, so I don't doubt someone out there has it in them.

 

You would have to know the 'field of view' of the lense and then the distance the camera was placed from the board to be able to reverse calculate the precise length due to perspective barrelling.

 

Doesn't have to be that precise