R600 and Xbox2 VPU joint effort by ArtX-ATI-Real3D teams?

Dave Orton
If you dissect in, for example, to the R600 product, with is our next, next generation, that development team is all three sites - Orlando, Silicon Valley, Marlborough – but the architectural centre team is in the Valley, as you point out, but all three are part of that organisation.

Orlando = former Lockheed Martin Real3D ?

Valley = ArtX / ATI West / Santa Clara, CA team

Marlborough = ATI's original team with probably some fresh engineers

(former Number 9 people are in there somewhere, too)

if indeed Xbox 2 is based on R600, and R600 is being made by all 3 ATI design teams, I expect some awesome things. I doubt Nvidia could match the combined efforts of former ArtX and Real3D engineers, not to mention Number 9 and ATI engineers also.
 
Uhh hmm they are the same people artx is no more its part of ati now.Ati got them back before they came out with the 9700. Why u think ati has been better for past 3 years.Im not sure what core xbox 2 will use tho i know the r500 got moved to r600.
 
Re: R600 and Xbox2 VPU joint effort by ArtX-ATI-Real3D teams

Megadrive1988 said:
if indeed Xbox 2 is based on R600, and R600 is being made by all 3 ATI design teams, I expect some awesome things. I doubt Nvidia could match the combined efforts of former ArtX and Real3D engineers, not to mention Number 9 and ATI engineers also.

I followed your reasoning, until the part with Nvidia...
Since that on the console side, nVIDIA seems only onboard of the... Phantom, next-gen.

;)

Btw, having a huge scale project being developed in 3 location (far from each others), is always a big risk when it comes to an industrie where time frames are so important.
And why Ati engineers must be that superior to nVIDIA engineers?
Only the ATI West proved theirselves as being really superior to the competition, we have to remember that their last card is really good on the performance side, but we also have to remember that's some kind of R300 under steroïds, with almost the same architecture, "refinied" as said Dave.

What would have been nice, it's if STI have choose the "fixed function" path, we could have compared the two architectures.
We'll see if an "excellent features set" (Xbox2/N5) can do better than an "flexible" architecture (PS3).
The "PS2 vs. the others" was a test (not really brilliant, i've to admit),
The "PS3 vs. the others" will be some kind of a showdown. (at least for the 3D enthusiasts, sony will continue to do what's more profitable to them)
 
well it seems to me that ATI may have an edge in the quality of their engineers. they have all of ArtX (former SGI) and they have at least some of Real3D (former Lockheed, GE Aerospace, Martin Marietta).

Nvidia has 3Dfx, Gigapixel and some SGI. I suppose Nvidia and ATI are about equal, but I'm slightly biased toward ArtX and Real3D so that means I lean slightly toward ATI.

anyway.....

I agree with the rest of your post. X2/N5 with exellent features vs PS3 with complete flexibility, will be an interesting showdown.

it'll be interesting to see if Nvidia has any hand at all in the Sony-designed
GPU in PS3. Nvidia could really benifit from getting into what is favored as the #1 next gen console. Sony could benifit from Nvidia's backend rendering pipeline. that's another topic though, and might very well be moot.
 
If the xbox 2 really wants to compete with sony in the next gen. Then MS and ATI should devolpe R600 which I think It could do 4 billion to 10 billion polygons raw.
 
Re: R600 and Xbox2 VPU joint effort by ArtX-ATI-Real3D teams

Megadrive1988 said:
if indeed Xbox 2 is based on R600, and R600 is being made by all 3 ATI design teams, I expect some awesome things. I doubt Nvidia could match the combined efforts of former ArtX and Real3D engineers, not to mention Number 9 and ATI engineers also.

Fanboyish conjecture. No bearing on reality...
 
Hey, what's up with all the lanboyish videocard manufacturer evangelism here lately? It's getting pretty nasty.
 
ahem, sorry for the video card / video card maker comments, they shouldn't really be part of the console forum.

keeping things strictly to console stuff, I think Dave Orton was as honest as he could be regarding ATI's roadmap. no detail given on what Xenon VPU actually is. I think it's either a hybrid of R500 and R600 as others have mentioned (for 2005) or a custom R600 or R600++ (for 2006)

course, I perfer the later option 8)
 
I don't think it is any of those after reading the interview. According to Dave R600 is being developed primarily by Santa Clara and Orlando with a little input from Marlborough due to them working on "special projects" and it won't be until R800 that all the teams are unified. The Marlborough team has another 18-24 months to work on the "special projects" and from the sound of it they are using a highly modified R400.


Just after the introduction of R300 you talked about R400, however shortly afterwards that appeared to go off the roadmap and R420 appeared – what happened during that period? And would that be related to upcoming contracts with console vendors?

We changed the roadmap.

You can look at cause and effect. That was not the cause, internal changes was the cause and the outcome was that we decided the best way to go forward was to do “thisâ€￾ with the PC roadmap and take “thatâ€￾ and use it with X-Box. But that wasn’t the cause of the roadmap change, we had to make this change anyhow due to execution issues.
 
My guesstimate :LOL: :

Well from the looks of things it looks as if xbox2 vpu, (no clue about Nintendos next vpu) will be BASED on the old R400, but really, these are just numbers. Whether it's R500 or R600 means little I guess, because it's just a custom chip that will only reside in xbox2. Obviously since it is coming out in a "R500-600" timeframe, it will have far more power than was originally intended for the old R400. Same basic architecture, but beefed up quite a bit.

So my final guess is this:
R500=HIGHLY modified and beefed up R300 (PS 3.0 added)
R600= OLD R400 which has become XGPU2 (both having DX10 support, mabye the xbox part only partial), only a little more powerfull than the xbox2 part because I think the PC part will come out later, and have of course a "Platinum Edition" option which obviously would be too costly for a console.

My humble take on things anyways.....it's always fun to look back and see how this pans out! :p
 
qwerty2000 said:
Then MS and ATI should devolpe R600 which I think It could do 4 billion to 10 billion polygons raw.

10 billion? :LOL: You must be out of your mind. ;) Maybe if the chip has 40 vertex shaders it'll do around 10 billion... Won't ever happen though.
 
Guden Oden said:
qwerty2000 said:
Then MS and ATI should devolpe R600 which I think It could do 4 billion to 10 billion polygons raw.

10 billion? :LOL: You must be out of your mind. ;) Maybe if the chip has 40 vertex shaders it'll do around 10 billion... Won't ever happen though.
Hmmm from what I have heard about the R400 based GPU, it only has "shaders", so I actually think that would be possible in theory. Of course, no one is going to make a game with all of the "shaders" being only vertex shaders, 10billion flat shaded polys would look kinda funny! (no sharp edges, but no texture?? or pixel shading either, kinda like a super duper Virtua Fighter 1 :LOL: )
 
BigGamer X said:
Guden Oden said:
qwerty2000 said:
Then MS and ATI should devolpe R600 which I think It could do 4 billion to 10 billion polygons raw.

10 billion? :LOL: You must be out of your mind. ;) Maybe if the chip has 40 vertex shaders it'll do around 10 billion... Won't ever happen though.
Hmmm from what I have heard about the R400 based GPU, it only has "shaders", so I actually think that would be possible in theory. Of course, no one is going to make a game with all of the "shaders" being only vertex shaders, 10billion flat shaded polys would look kinda funny! (no sharp edges, but no texture?? or pixel shading either, kinda like a super duper Virtua Fighter 1 :LOL: )


Or a super duper Jak and Dakster.:)
 
BigGamer X said:
Hmmm from what I have heard about the R400 based GPU, it only has "shaders"

You must have been listening to wild rumors then, as there is absolutely no concrete evidence to support something like that.

10billion flat shaded polys would look kinda funny! (no sharp edges, but no texture??

With 100+ million polys per frame packed into a TV-res screen, who needs textures or pixel shaders? It wouldn't be needed. Texturemapped polys are only there to hide the fact poly-pushing capabilities are currently very limited, if we could model all the details directly into the geometry, that would be a much better solution.
 
Guden Oden said:
BigGamer X said:
Hmmm from what I have heard about the R400 based GPU, it only has "shaders"

You must have been listening to wild rumors then, as there is absolutely no concrete evidence to support something like that.

I think he meant a unified shader scheme rather than a GPU with "just shaders." As for that Wavey has all but confirmed that whatever R400 is called now the chip will have configureable shaders that can handle both vertex and fragment processing.

Guden oden said:
biggamerx said:
10billion flat shaded polys would look kinda funny! (no sharp edges, but no texture??

With 100+ million polys per frame packed into a TV-res screen, who needs textures or pixel shaders? It wouldn't be needed. Texturemapped polys are only there to hide the fact poly-pushing capabilities are currently very limited, if we could model all the details directly into the geometry, that would be a much better solution.

Without a texture how else are you going to map a color onto a surface, among other things? You could "hard-wire" a color onto each vertex but this has it's own problems, among these compression becomes more difficult. I just don't see the advantage of coupling all of the possible surface characteristics (diff/spec color, shininess, reflectivity, opacity...) so tightly together. How would you handle an animated texture? Or a texture that moved like skin over a mesh?
 
Thank you that's what I meant, unified shader scheme, it's all new to me so I just went with "shaders" meaning one "shader" can do either vertex or pixel shading, am I right?
 
BigGamer X said:
Thank you that's what I meant, unified shader scheme, it's all new to me so I just went with "shaders" meaning one "shader" can do either vertex or pixel shading, am I right?

Yes, you are correct. Unified shading will allow the HW to dynamically load balance ALUs between vertex and fragment processing. So if the HW is running a stencil pass more resources can be used for vertex shading since little "pixel processing" is needed, and if the HW is rastering a large triangle the HW can send more resources towards fragment shading.
 
Guden Oden said:
With 100+ million polys per frame packed into a TV-res screen, who needs textures or pixel shaders? It wouldn't be needed. Texturemapped polys are only there to hide the fact poly-pushing capabilities are currently very limited, if we could model all the details directly into the geometry, that would be a much better solution.

You'd still need shaders for reflections, specular effects and all effects that can't really be done with geometry, including the fact that it would be easier to STILL have a texture to store colour information, even if the mesh is super-dense. the texture itself would be very simple, with the variations in colour provided by the variations in the geometry (no need for bump maps).

Unless u wanna go the ray-tracing route that is... ;)
 
Back
Top