there is no NV47, but G70 is on the way

[Geo]

Just spoke with someone benchmarking the g70's (and a lot of you will read his review tomorrow) that plainly said that g80 is nv50 just as g70 is nv47.

He also mentioned "three times shader performance" and that a "512" bit memory bus for g70ultra may not be far from the truth...

The plot thickens. Damnit, the dual-core hints just will not die. Not amd/intel "dual-core" perhaps, but something that after it has been thru a few "pass it ons" comes out as "dual-core" by the time it hits the public.

Edit: Tho come to think of it, it would be just like a marketing department to latch on to a publicly recognizable buzz phrase and ride it, even if it does violence to what it is publicly recognized for being. . .

 

[nutball]

The plot thickens. Damnit, the dual-core hints just will not die.

They will if people will let them.

 

[KimB]

Hopefully they will once the G70 is finally announced....

 

[DemoCoder]

Nope, cause then it will just be R520 dual-core and G80 dual core. Look at the silly R600 stuff floating around.

 

[Geo]

The plot thickens. Damnit, the dual-core hints just will not die.

They will if people will let them.

You have another explanation? Or are you suggesting Neliz is a bald-faced liar?

 

[_xxx_]

Here's something interesting to keep you going, from the console forum:

 

[_xxx_]

Oh, and:

 

[Jawed]

Shame there's no 16xAA numbers on there.

The 8x seems to perform really nicely though!

Jawed

 

[Jawed]

Euh, actually, 3DMk05 is a useless way of testing AA performance now I think about it - if the bench is vertex fetch limited then you've got spare AA power anyway.

Jawed

 

[KimB]

Well, also notice the lack of scaling with resolution. Just goes to show how little pixel-limited that benchmark is.

 

[Geo]

Yeah, another graph that screams "crank up the IQ already, for chrissakes!" at me.

 

[Jawed]

Only four hours sleep geo

Jawed

 

[Thowllly]

Yeah, another graph that screams "crank up the IQ already, for chrissakes!" at me.

Agreed. If nobody benchmarks at higher than 1600x1200 resolution I'm going to get mad. For almost five years I've been waiting for reviewers to start benchmarking at 2048x1536! (That's ever since I got my first 19" monitor). Let me reiterate

I've been waiting for five years for you reviewers to start benchmarking at 2048x1536! Five years! Why, oh why can't you start benchmarking at a decent resolution already...

 

[PatrickL]

I would be happy with 1900 1200

 

[HaLDoL]

Firingsquad benches at +16x12 resolutions

 

[Jawed]

Firing Squad benches at 2048x1536 sometimes.

Not a lot of games support that res.

Also, widescreen resolutions, such as 1920x1080 are more popular these days, but still not benched.

Jawed

 

[_xxx_]

Uhm, how many "top" games are still playable @2048? Not really meaningfull.

 

[N00b]

Uhm, how many "top" games are still playable @2048? Not really meaningfull.

How many "top" were still playable @1600x1200 three years ago? And still many reviews benched at that resolution. I'm all for 2048x1536 since increasingly more games tend to be cpu-bound at 1600x1200.

 

[Xmas]

I don't trust these results, at least for 8xS.

1280x1024 w/ 4xAA/8xAF (6398) is clearly slower than 1024x768 w/ 8xS/16xAF (7156)? That makes no sense as the latter should compute 20% more pixels.

 

[_xxx_]

That makes no sense as the latter should compute 20% more pixels.

Maybe they, umm, "optimized" something?