This is true????

This image is from GenTech and state the 460m will be more powerful than our GSODed 5870m:



Let's the holy/geek/fanboy war begins.........



PD: also it comes with 7 hours of SC2 for free

 

Hm. 7.4% increase. Obviously it should since it has more vRam but I would have thought 1.5 would be higher than 7.4%.

 

Looks more like that image is from nVidia and it's just posted on Kens site. It is well known why Fermi does a better job on DX11 benchmarks like those.
But heh, I could care less about the fanboy war

Q

 

Ken has said a few times that the 4xx cards are slightly better at dx11 benches than the 5870, but the 5870 pulls ahead in dx9/10/10.1

Anyway dx11 is still barely playable on either...

 

You can turn tesselation and most of DX11 off on games right? So the 5870 would be playable?

 

No fanboy war here. We already knew it would be stronger in Tessellation. DX11 is far more than Tessellation and the 460M sucks in DirectCompute which is utilizing the GPGPU power, which for most game developers is more important. Now they can have the GPU do the work that would be difficult on CPU. While Tessellation is more visible and easy to see, but games like BC2 are DX11 fully, but no tessellation.

And saying more powerful is wrong. Far from it. HD5870M has 1,200 gflops of power and the 460M probably around 400 gflops or less.

The tessellation performance of HD5870M is not good enough for all out gaming. But neither is it for 460M.

Example:
OMG the Honda Civic is 7% faster than a Chevy Aveo. No one cares because no one is going to ever race these...

Yes, almost all games can turn off tessellation. But you may want to look up DX11. DX11 is far more than tessellation and IMO tessellation isn't the most important or exciting feature either.

DirectCompute will offer better shadowing, better lighting, better environments like waves and fractals. IMO these things are more important than some extra polygons on a texture and often hard to tell if it's better or not.

http://www.theregister.co.uk/2009/09/28/microsoft_nvidia_collaboration/
This article explains a bit, but a bit fubar since it focuses on Nvidia CUDA which ironically perform terrible, horrible and downright awful for DirectCompute. Like HD5870 blowing a GTX 480 out of the water 3 times.

 

well it is a little better at tessellation as everyone has said. PLUS those numbers are wrong... i get 600 point in unigine heaven 2.0 with my stock clocks and default benchmarking settings, not 561 like the poster says.

 

well, the results advertised are somehow biased anyway since Nvidia makes customer think that their the best. Like the battle between Apple and Microsoft maybe. LOL. But according to results in some of the overclocking forums, Nvidia gets beaten by ATI when their matched with the same specs. Like the comparison between the GTX 480 (2GB) and RHD 5870 (1GB), the 480 outperformed the 5870, but when compared to the 5870 (2GB), the 480 was blown away even more when compared with the 5970 (2GB) & (4GB)

 

RAM buffer has almost 0 to do with GPU performance. It's the equivalent to saying cars with bigger gas tanks perform better than cars with smaller gas tanks.