R580 on B3D?

only thing i noticed different was the memory bus width being 64x4 rather then 32x8, not to say thats a typo though cause i dont know.
 
I've noticed the date is likely to be a bit wrong too....tsk tsk Dave, US was right then - most of it is wrong!
 
Unknown Soldier said:
:p

Just thought he'd want to remove it since it hasn't been released yet.

Hmm... maybe the date is about the time HE received the card? (and you suddenly linking to it seems like you're in the know too and just want a good laugh with everyone.)

Or maybe I'm too optimistic ;)
 
It's very likely that reviewers are already in possession (or close to) of a r580 if ati wanna launch it end of january.
 
tEd said:
It's very likely that reviewers are already in possession (or close to) of a r580 if ati wanna launch it end of january.

Last week if I correctly understood the seagull taking a dump on my window...

It is at those times that people "in the know" start doing things like this.. maybe under pressure of an NDA? who knows..
 
tEd said:
It's very likely that reviewers are already in possession (or close to) of a r580 if ati wanna launch it end of january.

For a Jan 24 launch? Only if they took Rys "most discouraging trend of 2005" to heart. But I'm thinking that's a little too much to hope for on the "quick service" front. :LOL:
 
geo said:
Only if they took Rys "most discouraging trend of 2005" to heart.

What, you mean this discouraging trend? :LOL:

No GDDR4 mentioned either. I make R580 about 18.5 x 19.5 ~= 360mm²; one might crudely extrapolate a transistor count of roughly 400 million. That sound about right?
 
MuFu said:
What, you mean this discouraging trend? :LOL:

No GDDR4 mentioned either. I make R580 about 18.5 x 19.5 ~= 360mm²; one might crudely extrapolate a transistor count of roughly 400 million. That sound about right?

I've seen estimates at around 350, so i'd agree with your estimate...
 
MuFu said:
What, you mean this discouraging trend? :LOL:

No GDDR4 mentioned either. I make R580 about 18.5 x 19.5 ~= 360mm²; one might crudely extrapolate a transistor count of roughly 400 million. That sound about right?

I was a bare scosh under that when I did it, but yeah "neighborhood".
 
Erm, we already know R580's pipelines and shader units. Same 16 pipelines, but triple the fragment shader units to 48. And (possibly unfortunately) same 16 TMUs and ROPs. I guess the same 8 vertex shaders, too.
 
Fodder said:
Only 8? Seems a little meagre, given RV530 has 5 compared to RV515's 2.

Well what do you expect then, 20 vertex shaders just because R580 is "4x RV530"?
 
Last edited by a moderator:
Kaotik said:
Well what do you expect then, 20 vertex shaders just because R580 is "4x RV530"?


(ssst.. they might read this at l'inq)

I guess it's still 8 vertex shaders.. but I think people were expecting 10 or 12 actualyl..
 
Kaotik said:
Well what do you expect then, 20 vertex shaders just because R580 is "4x RV530"?
No, but given that R580 looks to be be significantly more powerful (at times) than R520, and not clocked much higher, I expect "more".
 
Last edited by a moderator:
ATI made X700 unusually powerful in the VS dept., too. I guess the point is that vertices don't scale with screen res, while pixels do? Or maybe that game engines aren't as variable/scalable with their vertices or vertex effects as they are with their pixel effects?

But, yeah, Fodder, I expected maybe 10 VSs, but I can't say why other than it sounds more balanced. :)
 
Pete said:
Erm, we already know R580's pipelines and shader units. Same 16 pipelines, but triple the fragment shader units to 48. And (possibly unfortunately) same 16 TMUs and ROPs. I guess the same 8 vertex shaders, too.

Oops, sorry, I meant for any future updates, i.e. R590 or whatever?
 
Back
Top