AMD: R9xx Speculation

[UniversalTruth]

Physics Test
Rigid body physics simulation with a large number of objects
This test runs at a fixed resolution regardless of the chosen preset

http://www.3dmark.com/benchmarks/


Probably it's my misunderstanding and this test has nothing to do with nVidia's PhysX.

 

[Kaotik]

Slightly off, but, do we expect Cayman to be able to run 3D Mark 2011 smoothly? And, one more thing. WTH does that stupid physics test in it? Is it strictly nVidia oriented or this physics is anything different?

FM doesn't approve GPU PhysX enabled Vantage results, so it's not exactly "nVidia oriented" - even if 11 used PhysX too it wouldn't mean it's nV oriented

 

[Mize]

Dang! A kilobuck for the commercial use version?

 

[digitalwanderer]

Slightly off, but, do we expect Cayman to be able to run 3D Mark 2011 smoothly?

If history is any indicator no cards should be able to run 3dm2k11 smoothly for at least a generation or two.

 

[fellix]

For tEh LuLz:

Sorry, couldn't resist!

 

[digitalwanderer]

For tEh LuLz:

I hereby promote you and award you this medal, that was BRILLIANT!

 

[rpg.314]

Awesome

 

[iMacmatician]

Haha, that's awesome!

 

[caveman-jim]

http://www.3dmark.com/benchmarks/


Probably it's my misunderstanding and this test has nothing to do with nVidia's PhysX.

Could be the death knell of PhysX, to be added in Vantage and removed in '11 - a Bullet to the head?

 

[R300King!]

So does this prove that the original 6990 slide we was faked?

So someone just made up a Power Point slide with what they wanted, took a slightly blurred photo at an angle with IS off to make it look like they were trying to get a sneak shot of some yet unrelease NDA info and then posted it up as real??? I can't believe it.

Also, I don't know how popular that 3DMark 11 is going to be. If you were some Video card manufacturer (other than MSI) would you seriously run this demo at some sort of convention or disply booth constantly running 3DMark11 showing your competitor's name all over it? Didn't think so.

 

[Arnold Beckenbauer]

This "original" 6990 slide?
http://www.3dcenter.org/blog/leonidas/die-leichtglaeubigkeit-gegenueber-praesentationsfolien

 

[EduardoS]

1000 transistors each? I don't think so.

What do you think an adders is?

 

[Rys]

I've left a smattering of discussion about the fake slides so that it's in the thread for posterity, but I've nuked the rest of the noise.

 

[Shtal]

This is what I expect for camanXT.

 

[chavvdarrr]

This is what I expect for camanXT.

975MHz for 6970? I don't expect more than 900

 

[Arnold Beckenbauer]

This is what I expect for camanXT.

15 SIMDs per "Graphics Engine"?

 

[simbus82]

15 SIMDs per "Graphics Engine"?

I do not think the slide will provide us the true number of SIMD.

If you look carefully there are "dot" instead of SIMD: obviously they would not yet reveal the true numbers of simd!

 

[Gipsel]

http://img193.imageshack.us/img193/4112/comaparison.jpg

This is what I expect for camanXT.

That is not consistent:

3400 GFlop/s / 1920 / 2 = 885 MHz

1850 MTri/s / 2 = 925 MHz

And you can forget about 1.5 GHz GDDR5 with just 6 GBit/s chips. It will be a bit lower (I guess 1.4 GHz maximum).

 

[DavidGraham]

New 3DMark 11 results :

HD5870CF ：X2350

GTX580 ：X1748
HD5870 ：X1234
HD6870 ：X1120
GTX 460 ：X923
HD6850 ：X899
HD5830 ：X845

Looking at GTX 460 results vs the HD 6850 , I would say that tessellation gave the NVIDIA card a tiny boost .

CPU:i7 960
Windows 7 x64

262.99
260.99
10.10

http://www.chiphell.com/thread-141469-1-1.html
​

 

[psolord]

If there is some connection between the resulting framerate and the xrating of 3dmark10, then it will be a sad experience for most users! :S

Thanks for the info!