512SP GTX 480 tested!

Sceptical.
Though not surprised regarding the results of the "review". However, i still think a retail release of this card should be done (if it's true). Just think of the performance under water..
 
I'm surprised people were thinking that 32 extra SPs would make a huge difference, these margins aren't dissimilar from the difference between the GTX 192 and 216, and proportionally there was actually a bit extra additional shader power there.

The only real shock here is the power consumption.
 
It's not clear in the article, but I'd imagine the extra 200W comes from the shaders and clock speed increase to 800MHz (as well as the voltage increase required to stabilize it). I bet a GTX 480 @ 800MHz also comes quite a bit more power too.
 
Rossi~ said:
Sceptical.
Though not surprised regarding the results of the "review". However, i still think a retail release of this card should be done (if it's true). Just think of the performance under water..
Click to expand...
Yeah "under water" preferably the atlantic :) Damn that thing is hot and power hungry
 
Mr. K6 said:
It's not clear in the article, but I'd imagine the extra 200W comes from the shaders and clock speed increase to 800MHz (as well as the voltage increase required to stabilize it). I bet a GTX 480 @ 800MHz also comes quite a bit more power too.
Click to expand...

I would agree but I have a hard time buy those specs to begin with. unless they have made some startling progress at 40nm I can't see how this would work. they would be so much better off sticking two GF104 on a board and going with that
 
The results at stock clocks seem to be unsurprising. 6% performance increase on average, and the number of addtional SPs isn't much higher than that.
 
Valset said:
I would agree but I have a hard time buy those specs to begin with. unless they have made some startling progress at 40nm I can't see how this would work. they would be so much better off sticking two GF104 on a board and going with that
Click to expand...
I think they need 28nm to make this chip feasible, unless their engineers have found a few miracles in the design. Until then, they'll have to be satisfied with GF104 variants.
NKDietrich said:
The results at stock clocks seem to be unsurprising. 6% performance increase on average, and the number of addtional SPs isn't much higher than that.
Click to expand...
Near linear scaling of performance is impressive in my book. At least it shows this architecture is sound and has plenty of scaling still left in it. Like I said though, right now it just isn't feasible and the GF 100-based cards are a bad joke.
 
Mr. K6 said:
I think they need 28nm to make this chip feasible, unless their engineers have found a few miracles in the design. Until then, they'll have to be satisfied with GF104 variants.
Near linear scaling of performance is impressive in my book. At least it shows this architecture is sound and has plenty of scaling still left in it. Like I said though, right now it just isn't feasible and the GF 100-based cards are a bad joke.
Click to expand...

I agree, but how much longer will it take to transition to 28nm?
 
You must log in or register to reply here.
Back
Top