Ok, I got blasted for mentioning this idea in the thread speaking about the Xbox 360 leaked specs on pg. 11 or 12 I think.
I just not gonna let it go . Call me crazy but I feel the R500 may in fact be a SM4.0 part instead of SM3.0++ as has been the predominant assumption to date. I'm not saying it's an official SM4.0 part but it could be for all intents and purposes.
I'm thinking that just like the Xbox used a custom OS and directx version the Xbox 360 can as well but in this case directx is actually WGF2.0 or the part of it that handles 3D rendering directly.
I got blasted for thinking WGF2.0 was complete or even a version of Longhorn etc was. This is understandable, but I'm not really saying all that. What I'm saying is that they could use a beta version of or the Kernel that's going to end up in Longhorn and some beta/custom WGF2.0 'like' interface (where like could be beta or stop gap...pick what suits you...or don't). The Kernel would be the first thing anyone would get done in implementing an OS. It's not to hard for me to think Longhorn's is good to go beyond some tweaking with system level apps not yet complete but also not needed for the Xbox 360. WGF may not be complete in removing all the overhead in directx as it is now or having all parts of spec finalized but it's not so hard to think they got enough to code up a custom version ONLY for the Xbox 360's R500 VPU.(where any reduced overhead is still better than directx as it is now) I'd actually think getting WGF going would be an utter priority at MS given how Longhorn biggest draw is supposed to be a 3D desktop.
ATI also has been working on this VPU for how long? ...a long time. The R500 architecture has always been thought of as destined for SM4.0 compliance but limited to SM3.0++ in the Xbox 360. Well...why couldn't MS and ATI remove the limits if they already had stuff ready to go beyond the very same? Well maybe not ready to go "big time" but there in substance.
To me it seems like a great opportunity to experiment for both parties. ATI gets a spin with SM4.0 so that they can better tweak the R600 iteration and not only that whisper in MS's ear about things that gave them difficulty or they'd like to see in the final spec which would arrive later on PC. MS gets to see just which parts of the spec need revision, more emphasis, removal or inclusion. With performance data it be a great way of finding the bottleneck in their implementation and further eliminating the API's overhead. MS also gets two other advantages. It's get the jump on the OpenGL consortium which must answer WGF2.0's SM4.0 and all the other competition in console arena as they would surely be restricted to SM3.0++ (thus no load balancing on the GPU/VPU) unless they elected to craft an comparable API themselves (probability 0.0001 % but I'm just guessing it's hard to do this).
The latter is actually pretty significant with respect to the concept of "load balancing" on the VPU that I'm sure many here more capable than I are well aware of. This could be the reason why EA and others have recently commented to the affect the the Xbox 360 was more powerful than they expected (not to discount the GFLOP bump in the CPU of course).
Enough yammering I've made my case and if I'm wrong then I'm wrong but apart from my observations about the use of unified shaders found in the leaked specs and now the official specs here is an interesting comment that lends further credence to the idea that I am not completely insane
http://features.teamxbox.com/xbox/1145/The-Xbox-360-Dissected/p3/
I note this unified shader language and the mention of WGF2.0. This language would be the successor to HLSL and would compile up to what shader profile with WGF2.0...
I got excited here's another quote to this affect.
What say you? (Be gentle...)
I just not gonna let it go . Call me crazy but I feel the R500 may in fact be a SM4.0 part instead of SM3.0++ as has been the predominant assumption to date. I'm not saying it's an official SM4.0 part but it could be for all intents and purposes.
I'm thinking that just like the Xbox used a custom OS and directx version the Xbox 360 can as well but in this case directx is actually WGF2.0 or the part of it that handles 3D rendering directly.
I got blasted for thinking WGF2.0 was complete or even a version of Longhorn etc was. This is understandable, but I'm not really saying all that. What I'm saying is that they could use a beta version of or the Kernel that's going to end up in Longhorn and some beta/custom WGF2.0 'like' interface (where like could be beta or stop gap...pick what suits you...or don't). The Kernel would be the first thing anyone would get done in implementing an OS. It's not to hard for me to think Longhorn's is good to go beyond some tweaking with system level apps not yet complete but also not needed for the Xbox 360. WGF may not be complete in removing all the overhead in directx as it is now or having all parts of spec finalized but it's not so hard to think they got enough to code up a custom version ONLY for the Xbox 360's R500 VPU.(where any reduced overhead is still better than directx as it is now) I'd actually think getting WGF going would be an utter priority at MS given how Longhorn biggest draw is supposed to be a 3D desktop.
ATI also has been working on this VPU for how long? ...a long time. The R500 architecture has always been thought of as destined for SM4.0 compliance but limited to SM3.0++ in the Xbox 360. Well...why couldn't MS and ATI remove the limits if they already had stuff ready to go beyond the very same? Well maybe not ready to go "big time" but there in substance.
To me it seems like a great opportunity to experiment for both parties. ATI gets a spin with SM4.0 so that they can better tweak the R600 iteration and not only that whisper in MS's ear about things that gave them difficulty or they'd like to see in the final spec which would arrive later on PC. MS gets to see just which parts of the spec need revision, more emphasis, removal or inclusion. With performance data it be a great way of finding the bottleneck in their implementation and further eliminating the API's overhead. MS also gets two other advantages. It's get the jump on the OpenGL consortium which must answer WGF2.0's SM4.0 and all the other competition in console arena as they would surely be restricted to SM3.0++ (thus no load balancing on the GPU/VPU) unless they elected to craft an comparable API themselves (probability 0.0001 % but I'm just guessing it's hard to do this).
The latter is actually pretty significant with respect to the concept of "load balancing" on the VPU that I'm sure many here more capable than I are well aware of. This could be the reason why EA and others have recently commented to the affect the the Xbox 360 was more powerful than they expected (not to discount the GFLOP bump in the CPU of course).
Enough yammering I've made my case and if I'm wrong then I'm wrong but apart from my observations about the use of unified shaders found in the leaked specs and now the official specs here is an interesting comment that lends further credence to the idea that I am not completely insane
What sets the Xbox 360 apart from any other existing gaming hardware and puts the console at the very edge of real-time visuals is its graphics system. ATI Technologies, the partner selected to create the Xbox 360 GPU, has designed the very first graphics processor that offers a unified shader pipeline at a hardware level.
This hardware solution together with a unified shader language, which will debut on PCs in 2006 with the next version of Windows (code-named Longhorn) and in its Windows Graphics Foundation 2.0, put the Xbox 360 hardware at the forefront of graphic technologies.[/
http://features.teamxbox.com/xbox/1145/The-Xbox-360-Dissected/p3/
I note this unified shader language and the mention of WGF2.0. This language would be the successor to HLSL and would compile up to what shader profile with WGF2.0...
I got excited here's another quote to this affect.
With the advent of the next-generation graphics chips from ATI, such as the one used in the Xbox 360, and the next-generation of DirectX, which will built in the next-generation Windows operating system from Microsoft, (code-named Longhorn) as Windows Graphics Foundation 2.0, a unified shader model will become a reality. The PC industry will have to wait until late 2006. With the Xbox 360, we’ll get this later this year.
What say you? (Be gentle...)