Whew it's chilly in here. Did someone leave the Vapor-X open? Brrrr :

http://www.neoseeker.com/Articles/Hardware/Reviews/5870vaporx/

Daayyyuum, the pure benchmarks (Furmark and Blunderbuss), the 5870 rapes the competition.

That's the next gen boost for you TBH.

The problem with gaming benchmarks, especially comparing old and new gen, is that it doesn't take into account how much of an improvement it would bring, visually, it just goes by pure performance. THe 5870 is no slouch in the FPS category (usually beating everything but the 295), but the improvements, visually, are not taken into account in those tests.

BTW: Is it just me, or shouldn't the NEW generation of cards have more interface memory than the old? the BFG 295 has way more than any of the others? Granted, I don't really know what it's for, but it strikes me that if an older gen card can have so much, why can't a new one? More is better, I'd think, unless it provides no noticable benefit and increases the cost...

quote tallteen86

BTW: Is it just me, or shouldn't the NEW generation of cards have more interface memory than the old? the BFG 295 has way more than any of the others? Granted, I don't really know what it's for, but it strikes me that if an older gen card can have so much, why can't a new one? More is better, I'd think, unless it provides no noticable benefit and increases the cost...

Is that after you take account of it being split across 2GPUs for the GTX295?

The 5870 has plenty of memory bandwidth for a single GPU, and 1GB handles most games at 2560x1600. Of course, 2GB models will follow in time, as will an X2 (2GPU) version.

Even split between two, it is still on the top (I think another card also had over 400 too).

I'm not talking about the GPU RAM memory, I'm talking about "Interface Memory". It's in the 'bits' size range.

Thanks for clarifying that about the interface.

You're correct that using a wider interface costs more, partly because of the number of traces to the GPU. It's also difficult to connect so many traces for a wide interface to the smallest GPUs. AMD/ATI seem happy to stick with 256bit on their high end, and get sufficient bandwidth by using the fastest DDR5 (effectively twice the clock speed of GDDR3).