Why did Sony abandon the idea of having cell in PS3 doing gfx?

Not for any one building; but for any given time the upper-bound on data requirements is on the order of xxx MB of data, composed of variable amounts of both unique and repetitious content, in memory. It's not hard to imaging this quickly inflating to levels well beyond what a DVD-9 can support in a free-choice open world with a high degree of unique content.

 

how you handling the ps3 news vince bro? you had big hopes man.

ah fresh in my memory the past years we talk of these situations. irc you the one spread this cell gpu FUD? i think so. fun times.

 

Panajev2001a,

So what you are really saying is that since Cell is not powerful enough to procedurally create such realistic looking graphics they needed to prerender it.. And as such, in light of Sony's reluctance in using the modern VC1, they needed the bloated BD to store it.

[/bait]

 

That's just silly. If that little scene takes up hundreds of megabytes, then someone is doing something seriously wrong. Really, there is nothing in that scene (they don't even move around, it's basically an enclosed space without even too many textures!) that would take hundreds of megabytes unless they're using uncompressed textures, which is useless.

 

I really think that if Getaway was rendered on Cell then it would have been mentioned at the conference. Looking at interviews like this, I could see how some PS3 lovers could misinterpret it as the Getaway being rendered entirely on Cell, but that's not what he's saying.

My guess is there was some miscommunication somewhere. In that Chatani interview that Panajev2001a linked to, it's actually more grammatically sound if you remove the "not" in the second last sentence. Or maybe there was some misunderstanding by Chatani himself, as he didn't write the demo. The texture/lighting is probably what he was told, and it's referring to post-processing, as ShootMyMonkey suggested. That's the best kind of processing for Cell, because it's point sampled 4x32bit, there's lots of reuse between neighbouring pixels when doing blur (cache helps), and it's insanely parallel without dependencies.

 

Because they had to put two cells in each PS3, one for game engine and the other for graphics functions. Too expensive for console