• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

VGleaks: Orbis Unveiled! [Updated]

WarLox

Member

YES!

bpABFgt.gif
 
That "balanced for 14 CUs" sounds mighty suspicious. Maybe that's why some people claimed that "it's a wash" between Durante and Orbis? What does "balanced for 14 CUS" mean?
 

Mononoke

Banned
Maybe a dumb question (and I apologize if it's been asked to death)...

But how does this compare to current high end PC builds? I'm looking at that 3.5gb dedicated to games, and weeping at my 2gb GTX680 sli.

I wonder if this gen will push dedicated memory past 2gb. I know Max Payne did with the high res textures.
 

Takuan

Member
Not even gonna pretend to understand what those numbers and acronyms mean, but people seem happy, so hurray! Can we see the games now?
 

gofreak

GAF's Bob Woodward
Guys, a GCN 'render backend' has 4 ROPs.

8 render backends * 4 ROPs = 32 ROPs.

The 7950 has 8 'render backends' also, for example.
 

onQ123

Member
only thing that I see that I haven't really seen before is


'Display ScanOut Engine'


so is this like a scaler chip?
 

Tobor

Member
so its basically identical to durango? only different being ram set up, of which both have advantages and disadvantages

Exactly as it should be. Next gen multiplat development will pick right up where the last gen leaves off.

PC/360/PS3 Lol, Wii ---> PC/Durango/Orbis Lol, WiiU.
 

Pistolero

Member
The latest paper specs make the two consoles way closer than people initially suggested. In fact, if the extra hardware thrown in Durango works as intended, it will probably end up on top.
 
I hope we don't get something like the PS3 demo. That was underwhelming....

You mean Vision GT? I doubt it'd be like that. Kaz has said that the cars that they actually built cars for GT5 that are more suited for the PS4 than the PS3. It just comes down to the courses after that. Even limiting it to one course wouldn't be a bad thing. With GT2000 they ended up only using a Special Route course for its unveiling, and even eventually started showing of Seattle.
 
Memory:

4 GB unified system memory, 176 GB/s
3.5 available to games (estimate)


Nice my source was right.
I thought for sure we were getting Durango info but this is great. Excellent specs and its good that they are more or less confirmed.

The hardware is there now we just have to wait for the games.
 

Polari

Member
Sounds like a beast to me. When you look at the best the PS3 has given us and consider how much more powerful this is, I'm betting we'll see some pretty impressive results long-term.
 

i-Lo

Member
Guys, a GCN 'render backend' has 4 ROPs.

8 render backends * 4 ROPs = 32 ROPs.

The 7950 has 8 'render backends' also, for example.

So each render backend comprises for 4 ROPs?

And what about the texture unit count? That is a staggering 3.56 times lower than the base Pitcairn.
 
That "balanced for 14 CUs" sounds mighty suspicious. Maybe that's why some people claimed that "it's a wash" between Durante and Orbis? What does "balanced for 14 CUS" mean?

Maybe what proelite alluded to. Extra CUs on the GPU that will assist the CPU at time to time. IDK hopefully someone clarifies.
 
R

Rösti

Unconfirmed Member
Directx 11.1?
Compliant features. I will quote this from Microsoft's Dev Center site:

The following functionality has been added in Direct3D 11.1, which is included with Windows 8, Windows RT, and Windows Server 2012.
  • Shader tracing and compiler enhancements
  • Direct3D device sharing
  • Check support of new Direct3D 11.1 features and formats
  • Use HLSL minimum precision
  • Specify user clip planes in HLSL on feature level 9 and higher
  • Create larger constant buffers than a shader can access
  • Use logical operations in a render target
  • Force the sample count to create a rasterizer state
  • Process video resources with shaders
  • Extended support for shared Texture2D resources
  • Change subresources with new copy options
  • Discard resources and resource views
  • Support a larger number of UAVs
  • Bind a subrange of a constant buffer to a shader
  • Retrieve the subrange of a constant buffer that is bound to a shader
  • Clear all or part of a resource view
  • Map SRVs of dynamic buffers with NO_OVERWRITE
  • Use UAVs at every pipeline stage
  • Extended support for WARP devices
  • Use Direct3D in Session 0 processes
Source: http://msdn.microsoft.com/en-us/library/windows/desktop/hh404562(v=vs.85).aspx

This, HLSL 5 and the Compute Shader (DirectCompute) features on 11.x hardware is probably what they have been looking at in an API context. Apart from their own proprietary technology of course.

To comment on this "article", I'm moderately pleased out of a development standpoint (the I/D caches could be bigger though, 32KB is very common). But I don't get why they don't write the memory type (which is most likely GDDR5). Certainly what is communicated here doesn't suffice to make any greater statement about the system's capabilities, but overall it appears quite pleasant.
 

Vinci

Danish
Yeah, all that's about as informative for me as someone giving directions in Hindi. I'll just have to wait and see how they communicate it later on.
 
Top Bottom