Xenon Development Kits available.... NOW ( Alpha version )

Well apparently the word in the streets ( good streets ) of the Informatic High-way is that Xbox 2: ( code-name Xenon ) Development Kit in Alpha version are coming out... NOW.

Yes, it seems that for the elected few who have received notification and have agreements with Microsoft, they can download the Alpha XDK ( Xenon Development Kit ) as we speak.

The next revision of the SDK should be out around the first or second week of April.









Discuss.
















Oh, there is e-DRAM on the GPU: it should be for textures ( this is the voice going around ) like on GCN's Flipper chip ( the Texture Cache ), but I do not know if they will also store the back-buffer and Z-buffer in the e-DRAM.








Yes, the SDK give you a nice look of the machine specs and features... or so I hear.
 
Re: Xenon Development Kits available.... NOW ( Alpha version

Panajev2001a said:
Well apparently the word in the streets ( good streets ) of the Informatic High-way is that Xbox 2: ( code-name Xenon ) Development Kit in Alpha version are coming out... NOW.

Yes, it seems that for the elected few who have received notification and have agreements with Microsoft, they can download the Alpha XDK ( Xenon Development Kit ) as we speak.

The next revision of the SDK should be out around the first or second week of April.

Discuss.
Oh, there is e-DRAM on the GPU: it should be for textures ( this is the voice going around ) like on GCN's Flipper chip ( the Texture Cache ), but I do not know if they will also store the back-buffer and Z-buffer in the e-DRAM.

Yes, the SDK give you a nice look of the machine specs and features... or so I hear.

As I said Xbox is going to launch first! First time I got sg right :LOL: BTW, they sure like to copy Nintendo lately...
 
Re: Xenon Development Kits available.... NOW ( Alpha version

hupfinsgack said:
BTW, they sure like to copy Nintendo lately...

That's all they can do, just have a look at Windows, most of the ideas where introduced by Apple first.

Fredi
 
Is this legitamate? If so I'm willing to pay Visual Concepts a nice friendly visit and see what they have.
 
Joker said:
Is this for real?

two reliable "sources" are more or less confirming it, so yes it is...

BTW, is there anyone who remembers how much before the launch alpha-devkits were released last time? (discount the Xbox alphas as they were in rush to release them, so they're probably not very representative).
 
Oh, there is e-DRAM on the GPU: it should be for textures ( this is the voice going around ) like on GCN's Flipper chip ( the Texture Cache ), but I do not know if they will also store the back-buffer and Z-buffer in the e-DRAM.

This might be the part where the errors are...

The other part should be right as I heard in the streets that it is not an interpretation, but it is the word of the wise who knows the form of the Xenon and does not operate in the realm of opinions.
 
Panajev2001a said:
Oh, there is e-DRAM on the GPU: it should be for textures ( this is the voice going around ) like on GCN's Flipper chip ( the Texture Cache ), but I do not know if they will also store the back-buffer and Z-buffer in the e-DRAM.
This might be the part where the errors are...

Well, if so, either the RAM is not e-DRAM and something else or it also stores Z-buffer and back-buffer...
 
cthellis42 said:
DeanoC said:
Some of what Panajev wrote is wrong. ;)
You realize now people are going to be coming after you with torches and pitchforks, right? :p

mob.gif
 
I sure as hell hope there IS some type of embedded memory in Xbox 2's graphics processor.

it would help ease the burden on the reported 51 GB/sec main memory bandwidth (which sounds like alot today, but wont be in 2005-2006)
 
Going completely off-topic...

If you look at PC video cards you have two seperate paths (on a PC Video card, video RAM and AGP) it makes little sense to send all data over one of the path while the other sits idle. On PC its fairly common (or it was a few years ago) to send all vertex data over AGP, leaving the video RAM bandwidth purely for framebuffer writes and texture reads. The cost is amortilized due to a look ahead pre-shader vertex cache that can eliminate almost all stalls, even random access patterns are o.k. due to a hardware multi-threading architecture that usually has another thread to execute in the rare cache you do get a cache miss.

Now if you see texture reads and vertex reads as the same operations for future GPU's (there both just memory reads), it easy to see that a design that built on this idea would have 2 seperate paths to the GPU, one for read-only data (texture, vertices, indices etc) and one for write-only data (depth, colour). One wrinkle in this design is that a true segmented architecture would require a fast copy from write-only to read-only for render-to-texture/vertex/index array techniques.

Sorry for going completely off-topic :oops: :smile:
 
Back
Top