NVIDIA GF100 & Friends speculation

Because it's the 6970 crossfire the 580 SLI will be up against.

 

It's fair if they compare X1 mm^2 to X2 mm^2, where X1~ X2 and X1 is their GPU's area, and X2 is Radeon's GPU area.

 

I really hope that's not the final name…

 

No, SE stands for Slow Edition or Second Edition.

 

I guess it's just official higher clocked version so that they don't have to convince reviewers to use OC editions against Radeons anymore.

 

Guess again, it's a second (or third) bin SKU.

 

But it's clocked slightly lower and has fewer shaders enabled, according to Darth Shader's link. The fact that it'll be slower is typical of nV's use of SE naming, IIRC.

Any reviews showing how OCing the memory alone affects a 460? I'm curious if this 460SE can make up for its disabled shaders/TMUs with its greater bandwidth (both relative to the 460 768MB).

 

Ah yes, didn't see the slide.

 

Almost certainly this is slower than GTX460 768MB, unless you take some benchmarks into account which are limited by the amount of memory. There is only about a 10% difference between the GTX460 768MB and 1GB, I suspect losing the SM alone makes at least 10% difference, and then it has slower clock too (both chip and mem). So to get GTX460 768MB performance you'd probably need to overclock to normal GTX 460 clocks or very slightly more. If you want to reach GTX460 1GB performance, you'd likely need to OC the chip to around 750Mhz imho.
But as long as it's sold at GTX 460 768MB price level (and hence quite a bit cheaper than the GTX 460 1GB) it's not a bad deal imho.

 

How can this be compared to the leaked R6970 results? And does this benchmark scale well from quad to 6-core processor?
The Vantage result is with enabled PhysX.

 

They can be compare anything and someone will say it is fair, but since you give up potential when using dual GPU or Xfire/sli still I just ignore them. Give me a single GPU any day.

 

thats (only) 10 to 15% improvement over 480 GTX performance

 

Don't worry, there will be selected synthetic benchmarks where the difference is bigger!

 

75C minimum?

 

@30 FPS Avg.

a 725Mhz 480 does 38 fps avg. and a 5770 ~19FPS or something like that. looks like a cooked benchmark to me.

 

Amen. I'm a single GPU fan myself.

 

Congratulations sir

 

Maybe it's broken?

/runs away from neliz

Bastard!

hehe

 

Ah correct. Had a long time to see how Furmark fairs in framerate terms.

My 8800GT at the office runs at around 40fps at 1920X1080 noAA, so yes, there's probably something fishy here.