ATI R520 Details (not sure if true).. is this XBOX 2??

Qroach said:
Ask yourself, what's the difference between 600 million polys per second and 1.2 billion polys? IMO, a whole lot of really small polys you likely won't see.


Frame rate and large environments :p
 
frame rate is a good answer.

The amount of polys you throw at something won't make the environment larger. That's typically dependant on RAM and LOD.
 
Guden Oden said:
I don't expect the GPU in nextbox to beat the fastest available PC chip at the time of launch, there's basically no way in hell that could be possible, and I can't help but think anyone who does expect that is setting themselves up for (bitter) disappointment.

It'll be a nice powerful all-round package that through tighter integration and uniform hardware will produce better than PC-average gfx for a while like when the original xbox was released, perhaps for a longer time period this time as games are still lagging significantly behind the cutting edge of PC hardware... However, I can't for my life see how nextbox's GPU could be laying the smack down on PC from a raw performance standpoint. That doesn't make sense.

I think it could be as powerful as the most powerful pc chip, chips are pretty cheap to produce. Look at pc video cards, the first thing that is cut on the low end cards is memory, not the core. Xbox 2 will probably have a state of the art gpu, but with much slower memory.(And perhaps less of it)


And with this possibility comes the possibility of GT5 cars being even dumber then GT3's, fighters that run off of cliffs, and team mates acting like bees in a jar...
While it could help the PS3 it could also majorly hamper its AI and physics ability.

Well, that's what online gaming is for.
 
Fox5 said:
Well, that's what online gaming is for.

Broad band penetration is still poor for a mass market device that relies on online components for a good experience and last mile technologies are still not in full deployment. Not to mention a lot of people who can afford to live outside of big cities still have poor dialup ISPs. I know, I was one of them for many many years.

As much as I love online gaming (and as much as I dispise most single player games... HL2 is one of the first sp games I have enjoyed in a while) I think a lot needs to be done before online gaming can really be a real substitute in the mass market for poor/non-extent single player. Personally I think it will never happen because non-online games have a lot of genres and gametypes/modes that cannot be done online. Humans are not substitutes for epic stories with good gameplay.

That does not even begin to broach the issue of cheaters, hackers, and the onslaught of retards that make online play not so fun for many people. And online play is most fun for those good at it...
 
Acert93 said:
And online play is most fun for those good at it...

Bingo!

I don't dare go online anymore in Halo 2 or MechAssault 2 or UC. It's a death trap for me, and I'd rather have fun than get frustrated.
 
persiannight said:
http://www.neowin.net/forum/index.php?showtopic=284495

24 "Pipelines"
32 Texture Units
96 Arithmetic Logic Units (ALU)
192 Shader Operations per Cycle
700MHz Core
134.4 Billion Shader Operations per Second (at 700MHz)
256-bit 512MB 1.8GHz GDDR3 Memory
57.6 GB/sec Bandwidth (at 1.8GHz)
300-350 Million Transistors
90nm Manufacturing
Shader Model 3.0
ATI HyperMemory
ATI Multi Rendering Technology (AMR)
Launch: Q2 2005
Performance: Over 3x Radeon X800 XT !!! (for single R520)
16x stochastic FSAA
FP32 blending, texturing
Programmable Primitive Processor/Tesselator


nope
 
Qroach said:
How exactly? both graphic chips will be capable of rendering so many polygons, how will it make a difference??

I think its safe to say that the amount of polygons when comparing X2 and PS3 won't make a lick of difference. It's going to come down to shading and texturing, and both of GPU's will be handling 100% of that.

I wasn't talking about geometry. Consider this scenario. All vertex ops done on the Cell CPU, the GPU is dedicated solely to pixel shading. Imagine an entire PC card today being dedicated to pixel shading. Of course, there's the question of how hard the CPU would be pushed to feed the graphics card. However I don't think it'd require all 8 SPEs working on vertices...it'd be up to the developer what sort of tradeoff they would make. But regardless of geometry, they would know they'd have an ass-load of pixel shading power at the other end that can be used without any compromises.

Consider the rumoured spec of Xbox2. A unified shading pipeline and the possibility of the CPU being used for vertex ops. In order to compete with the imagined PS3 configuration above, on paper at least, you'd have to dedicate all your graphics shading units to pixels, and use your CPU for vertex ops. Which CPU to you fancy would come off better in that matchup, and having power to spare for the actual game? My money's on PS3's. Even the Xbox2 CPU could feed its GPU at the same kind of level as PS3, and even if the Xbox2's GPU had a total number of shading units equivalent to PS3's number of pixel shaders, there'd be the question of how good they'd be compared to PS3's i.e. a more general unit versus a dedicated unit (NVidia believes you would lose efficiency etc. with a unified pipeline). I'm assuming in this little scenario that both system's pixel shading instruction set etc. are roughly equivalent.

In summary, as I see it, the PS3 CPU is in a very good position to free up its GPU from vertex ops and allow it to focus on pixel shading - there's a lot of power there, arguably power to spare, and the SPE's would make good vertex shaders. Looking at the leaked Xbox2 specs, I think it'd have a hard time matching PS3's graphics performance if they did let the CPU deal with vertex ops and dedicated to the GPU to pixel shading (and again, this isn't so much about geometry, but simply the level of pixel shading that could be used if the GPU only had it to worry about!).

Again, that's just one scenario, of many possibilities..perhaps I'm making some inaccurate assumptions here..?
 
Qroach said:
How exactly? both graphic chips will be capable of rendering so many polygons, how will it make a difference??

I think its safe to say that the amount of polygons when comparing X2 and PS3 won't make a lick of difference. It's going to come down to shading and texturing, and both of GPU's will be handling 100% of that.

Actually not even on shading capabilities or textures.
Next gen graphics will be so saturated with polygons, shaders and hi-res textures that in the end the only thing that will make a difference is how things move. And maybe how many things are on screen at once.
 
Yes, if xbox2 games manage to look superior to PS3 and Revolution games, it has all to do with just art, not at all with more powerful technology, right?
 
PC-Engine said:
It'll come down to art as far as visuals are concerned.

Like in movies. A DVD looks real, what makes it look good is the art, design, cinematography, camera direction. All those little things will be much more important in the next years.
 
Next gen games will be far far far from "real", thus technological diferences will have impact on the visual side of things too.
We might not see them so clearly in first gen titles, partly because we'll be "awestruck" by the new detail and partly because first gen titles will be more affected by art.
But in the end, we'll see, I think.

You can see the difference between Shrek 2, and some other 3D animation that was done with lesser money, art- and technical resourses.
Why wouldn't you be able to see the differences between next gen consoles then. You'll just have to adapt and become educated where to see them.

Art will become more important, and bad art will be more visible next gen. But art scales hand in hand with technology, and if there are the rumoured differences between next gen, we will see what differences are because of art and what because of tech.
 
london-boy said:
Actually not even on shading capabilities or textures.
Next gen graphics will be so saturated with polygons, shaders and hi-res textures that in the end the only thing that will make a difference is how things move. And maybe how many things are on screen at once.

We can at least dream of hi-res textures :( Hard to fit hi-res textures, normal maps, etc... into 256MB of total system memory. Consider: Resolutions are going to jumping from 480i/p in most cases to 720p. That is over a 300% increase in pixels. If you have ever increased a games resolution that has low-res textures it becomes more and more appearant as you raise the resolution. We are getting are probably getting 400% more memory on the new consoles compared to the Xbox (256MB vs 64MB), so a lot of that new memory will be spent just making the pixels in 720p look comparably clear. That is a good thing mind you since you end up with more detail, BUT you have to consider new games will have more AI, Physics, and general game logic code. Then add in that million vertex meshes take up more memory than 600 vertex meshes. Throw in all the textures to texture these and then add MORE objects to the screen (we all want more enemies and more interactive objects in our gaming worlds) and I see the 256MB being hit so fast it wont even be funny. First thing to go will be all those pretty hi-res textures :( Or all the props and details in the game :( Or <edag!> low res everything! Ok, it wont be quite as bad as I think but I am hoping one of the big three gets the guts to throw 512MB in ;)

Btw, you are dead on with animation. We often forget how important animation is to graphics. It is one thing to look nice standing still, but how does it perform when in motion? Some of the best looking screenshots have been from some of the worse playing games.

Seamless animation and realistic interaction with environments will be the key features I am looking toward next gen (everyone will have pretty graphics in 720p... with low rez textures muwahaha... just kidding). Unfortunately these seem to be the hardest areas to master. . .

Ok, enough silliness... Time for bed... @ 7am :|
 
Some of you guys are a bit thick headed aint ya! ;)
I wasn't saying next gen will look real, or that we will have "enough" high res textures, or even enough of anything else really (if we had enough of anything, there would be no point for technological progression after the next gen).

What i meant is that at parity of visuals (same levels of poly counts and textures and shaders and whatever else), like DVD to a different DVD (Eternal Sunshine compared to LOTR), like XBOX2 to PS3 (mostly, they will have obvious differences, but not anything strikingly different from each other IMO), what will set something apart from the crowd will be the Art and the Animation/Physics.

Art because we will have loads of Doom3 kind of games, where it's all shiny shine and crap designs, then other games like a potential next gen ICO. People will know what looks best.

Animation/Physics because our brain is much more susceptible to how things move than how detailed they look.
 
I agree. Make a game with realistic moving characters (I mean really realistic) displaying real human emotions and facial expressions, and people would be completely spooked out. Even if the rendering was sub-par to other engines around.

We're so used to seeing stupid zombie like characters, that we don't even notice anything wrong anymore.
 
I think one very important thing will be in fluid, cloth and hair dynamics. Terribly complicated to do right, but we nee dto get those right to have realistic characters.
One day (maybe not even in the next gen, but u never know) we'll have characters modelled "naked" (with a leaf there ;) )and then realistic clothes will be put over them, like in reality. And they will move realistically. And hopefully not magically run through surfaces.
 
HL2 is a good example of art direction making a games rendering technology appear to be more than it really is. Not that Source is a bad rendering engine--it is quite good comparitively. But HL2/Source is more about the physics and how all the pieces come together than just a top of the line, cutting edge renderer. But many people (and even some journalists) gave HL2 awards for the best looking game of 2004. Obviously certain games shine in specific graphics technologies (e.g. D3 and shadows/lighting, FarCry with very far view distances) but the art direction really matched the theme and was well executed in HL2.

The fact everything moved well (animation, facial expressions, physics based interaction with environment) didn't hurt either.
 
Personally, I am just looking forward to seeing better physics next generation. IMO the key to realistic graphics isn't the accuracy of the textures, but movement and realistic lighting. It would be a great start if we had flawless LOD approaches, since most things far away wouldn't require high-res textures anyway... (since they're too far away and if the sun would be coming from there, you wouldn't make out anything to beginn with)...
 
What are all these pixels shaders going to do? Textures, lighting effects, a few optical effects...how many do we actualy need?

How do current top-end PC games make use of pixel shaders? Is there much room for improvement?
 
Back
Top