A new Sony patent application..

[Simon F]

Gosh it was from PCX1 days and as for patents, I haven't the faintest idea.

 

[Ug Lee]

Having read the patent in its entirety I have a few questions for those willing to push a (not so) young padawan learner in the right direction.

With the setting up of the bricks, what is the point of having them overlap, or be configured in strange ways? Would it just be a problem caused by the grouping of the objects into bricks?

Distributed rendering - it sounds to me like something a little too ambitious for the current times. But if implemented it might be something that could be pulled off in a small geographical area like Japan, and particularly with MMORPGs. I also wonder if when Sony talk about PCs in the patent, they aren't also referring to PS3s set up as nodes. It sounds like something that would be very expensive to get running on a global scale.

Perhaps Sony are looking at this as something akin to treading water? A limited 'test' to see how it goes? Thoughts on that would be appreciated, as to me the whole thing sounds a little too 'Jetsons' at this point in time.

 

[pcostabel]

With the setting up of the bricks, what is the point of having them overlap, or be configured in strange ways? Would it just be a problem caused by the grouping of the objects into bricks?

Unless you want to split the geometry, bricks will always overlap.

Distributed rendering - it sounds to me like something a little too ambitious for the current times. But if implemented it might be something that could be pulled off in a small geographical area like Japan, and particularly with MMORPGs.

Read the patent. Rendering will be distributed based on the network topology: objects that don't ned to be updated often will render on remote nodes, while object closer to the player will update on local nodes.

I also wonder if when Sony talk about PCs in the patent, they aren't also referring to PS3s set up as nodes. It sounds like something that would be very expensive to get running on a global scale.

This architecture could be used for renderfarms for non real time rendering. Service provider could sell computing power the same way they sell bandwidth now. In order for this to happen, the ISA must be the same for all nodes. That's what Cell is all about: a consistent ISA scalable from consoles to workstations to supercomputer.

Perhaps Sony are looking at this as something akin to treading water? A limited 'test' to see how it goes? Thoughts on that would be appreciated, as to me the whole thing sounds a little too 'Jetsons' at this point in time.

If they want to replace the x86 ISA as the defacto standard, they need to reach market penetration before the bandwidth catches on. A 100 million PS3 is a good start to estabilish Cell. Couple that with a few million workstations in CG companies and a few Cell-based supercomputer and the 'Cell-anywhere' plan starts to make sense...